{"id":"src__1wW8sp0gv9nbxCNdXeXW","slug":"anthropic-news","name":"Anthropic News","type":"scrape","url":"https://www.anthropic.com/news","orgId":"org_HXldkKAl-bZUWW6lNnELP","productId":null,"productSlug":null,"org":{"id":"org_HXldkKAl-bZUWW6lNnELP","slug":"anthropic","name":"Anthropic"},"isPrimary":false,"isHidden":false,"discovery":"curated","fetchPriority":"paused","metadata":"{\"parseInstructions\":\"Corporate news page — mixes product launches with partnerships, policy, and funding announcements. HIGH NOISE but occasionally carries critical launches (e.g. Anthropic Labs, Claude Design, new consumer products) that don't appear on any other source. Every post MUST be evaluated individually.\\n\\nINCLUDE when the post announces something users or developers can now use or try:\\n- New model launches (Opus, Sonnet, Haiku, and variants)\\n- New products or apps (e.g. Anthropic Labs, Claude Design, Claude for X)\\n- Major features or capabilities users can access today\\n- New platform services, tools, or SDKs\\n- Significant expansions of existing products (e.g. GA of a tier, new region)\\n\\nSKIP when the post is primarily about:\\n- Partnerships, customer stories, government MOUs, policy statements\\n- Office openings, hiring announcements, exec moves\\n- Funding, investment news, acquisitions of other companies\\n- Research publications, safety reports, evaluation methodologies\\n- Opinion pieces, industry commentary, thought leadership\\n\\nWhen in doubt: if the post describes something a user or developer can USE or TRY, include it. Noise is acceptable; missing a real launch is not.\",\"noFeedFound\":true,\"pageContentHash\":\"1f894c3c8881e653d7213d9dd3963d00a138e837da3e4b2f5bd45d20a61c5beb\",\"crawlEnabled\":true,\"fetchMethod\":\"crawl\",\"renderRequired\":false,\"crawlRender\":false,\"lastCrawlJobId\":\"0ab56572-6930-4e24-be80-761b29656f3a\",\"lastCrawlAt\":\"2026-11-06T13:43:42.676Z\",\"marketingFilter\":true}","notice":null,"kind":"platform","stars":null,"starsFetchedAt":null,"releaseCount":53,"releasesLast30Days":8,"avgReleasesPerWeek":1.6,"latestVersion":null,"latestDate":"2026-10-27T00:00:00.000Z","changelogUrl":null,"hasChangelogFile":false,"lastFetchedAt":"2026-11-06T13:44:11.282Z","lastPolledAt":"2026-11-05T00:47:09.827Z","changeDetectedAt":null,"trackingSince":"2026-01-22T00:00:00.000Z","releases":[{"id":"rel_9N2KJq9et8LGXygmZc9AR","version":null,"type":"feature","title":"Early progress on accelerating drug development with Claude","summary":"Anthropic says its scientists have spent several months in deep research partnerships with six organizations developing new medicines, as part of a broader effort to speed up drug development with Claude.","titleGenerated":"Anthropic reports early progress on accelerating drug development with Claude","titleShort":"Early drug development progress with Claude","breaking":"none","importance":3,"content":"We're at the start of a major effort to speed up drug development with Claude. Over the past several months, our scientists have engaged in deep research partnerships with six organizations developing new medicines.","publishedAt":"2026-10-27T00:00:00.000Z","fetchedAt":"2026-10-31T03:57:25.022Z","url":"https://www.anthropic.com/research/early-progress-drug-development","media":[],"coverageCount":0},{"id":"rel_q8zbCCSf6X4p4Yvwc_8BQ","version":null,"type":"feature","title":"Donating Agent Skills to the Agentic AI Foundation","summary":"Anthropic is donating Agent Skills, the open standard giving AI agents specialized knowledge and workflows, to the Agentic AI Foundation. More than 70 agent platforms including ChatGPT, Cursor, GitHub Copilot, and VS Code have integrated the standard since it was published in December 2025.","titleGenerated":"Anthropic donates Agent Skills open standard to Agentic AI Foundation","titleShort":"Agent Skills standard donated to Agentic AI Foundation","breaking":"none","importance":4,"content":"We're donating Agent Skills, the open standard that gives AI agents specialized knowledge and workflows, to the Agentic AI Foundation. More than 70 agent platforms, including ChatGPT, Cursor, GitHub Copilot, and VS Code, have integrated Agent Skills since we published the standard in December 2025.","publishedAt":"2026-10-22T00:00:00.000Z","fetchedAt":"2026-10-23T12:48:54.276Z","url":"https://www.anthropic.com/news/donating-agent-skills-to-the-agentic-ai-foundation","media":[],"coverageCount":0},{"id":"rel_t0m76DI8SLyoBIjXRPCrs","version":null,"type":"feature","title":"Investing in the public benefit","summary":"Anthropic is committing a portion of its shares to the Anthropic Public Benefit Fund, a wholly-owned LLC, with 5% of shares committed today subject to stockholder approval and an additional 2% added at the board's discretion for every 50% increase in company value above a baseline. Combined with co-founder pledges of 80% of equity and the employee equity match program, roughly 20% of Anthropic shares will be dedicated to the public good; the Fund will not be used for lobbying or political giving.","titleGenerated":"Anthropic commits 5% of shares to new Public Benefit Fund","titleShort":"Anthropic commits 5% of shares to Public Benefit Fund","breaking":"none","importance":4,"content":"Anthropic is committing a portion of its shares to the Anthropic Public Benefit Fund, a wholly-owned LLC. Subject to stockholder approval, 5% of shares are committed today, with an additional 2% added at the board's discretion for every 50% increase in company value above a baseline. The Fund gives flexibility to pursue mission work in new ways (e.g., large nonprofit donations or direct distributions) and will not be used for lobbying or political giving. Combined with co-founder pledges (80% of equity) and the employee equity match program, roughly 20% of Anthropic shares will be dedicated to the public good. More details on structure, governance, and valuation milestones to come, plus an upcoming public benefit report.","publishedAt":"2026-10-18T00:00:00.000Z","fetchedAt":"2026-10-19T12:55:20.183Z","url":"https://www.anthropic.com/news/public-benefit-fund","media":[{"type":"image","url":"https://www-cdn.anthropic.com/images/4zrzovbb/website/cbdc2eac6dfb37cf596d8f9fb32783036b5737c2-2000x1125.jpg","r2Key":"releases/e65c4b2b4a225d01d0c43a5dbcb811107fca13914fcfffd9ba46fcaadd322734.jpg","r2Url":"https://media.releases.sh/releases/e65c4b2b4a225d01d0c43a5dbcb811107fca13914fcfffd9ba46fcaadd322734.jpg"}],"coverageCount":0},{"id":"rel_LDmdOtqbUW7Z89P24s2tZ","version":null,"type":"feature","title":"Preparing for the 2026 US midterms","summary":"Anthropic published an update on its preparations for the 2026 US midterms, following earlier election integrity preparations shared this year.","titleGenerated":"Anthropic shares election integrity preparations for 2026 US midterms","titleShort":"Preparations for the 2026 US midterms","breaking":"none","importance":2,"content":"Update on Anthropic's preparations for the 2026 US midterms, following earlier election integrity preparations shared this year.","publishedAt":"2026-10-14T00:00:00.000Z","fetchedAt":"2026-10-16T13:03:09.820Z","url":"https://www.anthropic.com/news/2026-us-midterms","media":[{"type":"image","url":"https://www-cdn.anthropic.com/images/4zrzovbb/website/609927c7af2a454cedad848e4b2adc980c1f5fe6-2000x1125.jpg"},{"type":"image","url":"https://www-cdn.anthropic.com/images/4zrzovbb/website/d0efef057e61d2c590d06d8902da2167fe93200a-1920x1080.png"},{"type":"image","url":"https://www-cdn.anthropic.com/images/4zrzovbb/website/7b638266c2b8de46e1732a5e471b9a95b79b052d-1920x1386.png"},{"type":"image","url":"https://www-cdn.anthropic.com/images/4zrzovbb/website/35c702ec42254d97fd11618ff3f5dcc6845d4612-1920x1834.png"}],"coverageCount":0},{"id":"rel_XtDLZa6bbWy51hNiikDAr","version":null,"type":"feature","title":"Strengthening our defenses against distillation attacks","summary":"Anthropic describes reducing the rate of illicit distillation, in which model developers copy frontier AI capabilities without creating proportionally robust safeguards, by more than 90%.","titleGenerated":"Anthropic cuts distillation attack rate by more than 90%","titleShort":"Distillation attack rate cut by more than 90%","breaking":"none","importance":4,"content":"Illicit distillation allows model developers to copy frontier AI capabilities without creating proportionally robust safeguards. Here, we describe how we've reduced the rate of distillation by more than 90%.","publishedAt":"2026-10-12T00:00:00.000Z","fetchedAt":"2026-10-31T03:57:25.622Z","url":"https://www.anthropic.com/news/distillation-defenses","media":[],"coverageCount":0},{"id":"rel_u_oYhZJ9s1ji8fnu0g-uK","version":null,"type":"feature","title":"Introducing the Anthropic Cyber Mission","summary":"Anthropic introduces its Cyber Mission.","titleGenerated":"Anthropic introduces the Anthropic Cyber Mission","titleShort":"Anthropic Cyber Mission introduced","breaking":"none","importance":3,"content":"Anthropic introduces its Cyber Mission.","publishedAt":"2026-10-08T00:00:00.000Z","fetchedAt":"2026-10-09T13:04:19.260Z","url":"https://www.anthropic.com/news/anthropic-cyber-mission","media":[{"type":"image","url":"https://www-cdn.anthropic.com/images/4zrzovbb/website/e6614df689675126bb32ceb5cfcaaae016050c36-2000x1125.jpg"}],"coverageCount":0},{"id":"rel_WQ1qsEMC-GexSBT-ElMyz","version":null,"type":"feature","title":"Building on our commitment to American scientific discovery","summary":"Anthropic announced commitments supporting American scientific discovery.","titleGenerated":"Anthropic announces commitments supporting American scientific discovery","titleShort":"Commitments to American scientific discovery","breaking":"none","importance":2,"content":"Anthropic announces commitments supporting American scientific discovery.","publishedAt":"2026-10-08T00:00:00.000Z","fetchedAt":"2026-10-09T13:04:19.260Z","url":"https://www.anthropic.com/news/genesis-mission-commitment","media":[{"type":"image","url":"https://www-cdn.anthropic.com/images/4zrzovbb/website/d6058e0db8e477dc782dacae46e2ec6663d165d9-1000x1000.svg"}],"coverageCount":0},{"id":"rel_bQNpAwWG4PFR9P_dkK3a-","version":null,"type":"feature","title":"2026 Usage Policy update","summary":"Anthropic updated its Usage Policy.","titleGenerated":"Anthropic updates its Usage Policy","titleShort":"Updated Usage Policy","breaking":"unknown","importance":2,"content":"Update to Anthropic's Usage Policy.","publishedAt":"2026-10-08T00:00:00.000Z","fetchedAt":"2026-10-09T13:04:19.222Z","url":"https://www.anthropic.com/news/2026-usage-policy-update","media":[],"coverageCount":0},{"id":"rel_npZB9g8w5y5XxOcYZeGCs","version":null,"type":"feature","title":"Introducing Claude Haiku 5.5","summary":"Anthropic introduced Claude Haiku 5.5, described as its fastest, cheapest, and most capable small model yet, designed for high-volume, cost-sensitive work.","titleGenerated":"Anthropic launches Claude Haiku 5.5, its fastest and cheapest small model","titleShort":"Claude Haiku 5.5 launches for high-volume, cost-sensitive work","breaking":"none","importance":4,"content":"Our fastest, cheapest, and most capable small model yet. It's designed for high-volume, cost-sensitive work.","publishedAt":"2026-10-07T00:00:00.000Z","fetchedAt":"2026-10-08T12:50:46.078Z","url":"https://www.anthropic.com/claude-haiku-5-5","media":[{"type":"image","url":"https://www-cdn.anthropic.com/images/4zrzovbb/website/2b1d3817fe9aa732e1be4ed18446a1b1367379e7-1920x2322.png","alt":"Claude Haiku 5.5 benchmark chart"}],"coverageCount":0},{"id":"rel_-DPQB9xJPXwUqpjGwB7du","version":null,"type":"feature","title":"Expanding the Cyber Verification Program","summary":"Anthropic launched an expanded version of its Cyber Verification Program, giving qualifying security professionals access to advanced cyber capabilities and reduced blocking classifiers.","titleGenerated":"Anthropic expands Cyber Verification Program for security professionals","titleShort":"Cyber Verification Program expanded","breaking":"none","importance":3,"content":"Launching a new, expanded version of the Cyber Verification Program, which makes advanced cyber capabilities and reduced blocking classifiers available to qualifying security professionals.","publishedAt":"2026-10-06T00:00:00.000Z","fetchedAt":"2026-10-08T12:50:46.078Z","url":"https://www.anthropic.com/news/cyber-verification-program","media":[{"type":"image","url":"https://www-cdn.anthropic.com/images/4zrzovbb/website/2b1d3817fe9aa732e1be4ed18446a1b1367379e7-1920x2322.png","alt":"Overview of the Cyber Verification Program tiers"},{"type":"image","url":"https://www-cdn.anthropic.com/images/4zrzovbb/website/bdb05797e26ba2d096ebda6df1c762518da39db2-1920x1080.png","alt":"CyScenarioBench safeguard blocking results"}],"coverageCount":0},{"id":"rel_RwQLn5jUSRG3XLkJSoTCu","version":"Claude Sonnet 5.5","type":"feature","title":"Introducing Claude Sonnet 5.5","summary":"Claude Sonnet 5.5 is the second model in the Claude 5.5 family, generating outputs 30%+ faster than Sonnet 5 and typically costing up to 30% less per task at unchanged token pricing. It scores 70.6% on Terminal-Bench 4.0 (up from 10.3%), comes within two points of Opus 5.5 on GDPval-AA and CursorBench, and is the first Sonnet model to launch with cyber safeguards comparable to Opus 5.","titleGenerated":"Anthropic launches Claude Sonnet 5.5 with 30% faster output and lower per-task cost","titleShort":"Claude Sonnet 5.5: 30% faster, up to 30% cheaper per task","breaking":"none","importance":5,"content":"Claude Sonnet 5.5 is the second model in the Claude 5.5 family. It's a clear upgrade over Claude Sonnet 5, runs 30%+ faster, and costs up to 30% less for most work.\n\n**Key improvements:**\n- **Performance**: Scores 70.6% on Terminal-Bench 4.0 (vs Sonnet 5's 10.3%), two points below Opus 5.5 on GDPval-AA, and is the first Sonnet model to beat Pokémon Red from screenshots alone.\n- **Collaboration**: Writes more clearly than previous generation; better partner for collaboration with improved speed for fast iteration.\n- **Cost**: Same pricing as Sonnet 5 ($2/M input tokens, $10/M output tokens, $0.20/M cache reads) but typically needs far fewer tokens, costing up to 30% less per task.\n- **Speed**: Generates outputs 30%+ faster than Sonnet 5.\n- **Alignment and safety**: Improves on or matches Sonnet 5 on most alignment measures. First Sonnet model to launch with cyber safeguards comparable to Opus 5. Biology safeguards match Sonnet 5.\n\n**Coding improvements**: Particularly strong gains in agentic coding. On FrontierCode at High effort, scores 10 points higher than Sonnet 5 at one-fifteenth the cost. On CursorBench, within two points of Opus 5.5. Early testers noted faster codebase understanding and improved efficiency with better tool call batching.\n\nClaude Haiku 5.5 will join the family in coming weeks.","publishedAt":"2026-09-28T00:00:00.000Z","fetchedAt":"2026-10-02T11:50:51.069Z","url":"https://www.anthropic.com/news#claude-sonnet-5-5","media":[],"coverageCount":0},{"id":"rel_qZtrMulm8uggbyK_en_0X","version":null,"type":"feature","title":"Claude discovers a novel enzyme system with CRISPR-like repeats","summary":"Anthropic's new life sciences research group reports that roughly 950 Claude agents searching over 200,000 reverse transcriptases identified array-associated reverse transcriptases (ART), an enzyme system in bacteriophages with DNA repeat arrays resembling CRISPR. The lab's first experiments show the ART array is expressed as distinct short RNAs, and a pre-print describing the findings has been released.","titleGenerated":"Anthropic says Claude discovered novel CRISPR-like enzyme system ART","titleShort":"Claude autonomously discovers CRISPR-like ART enzyme system","breaking":"none","importance":4,"content":"We're introducing a new life sciences research group and laboratory at Anthropic. Our focus is on fundamental biology research using Claude: exploring datasets of DNA to identify uncharacterized protein families, generating hypotheses at scale, and testing them through experiments in the lab. This post introduces the team behind this work and shares early results in which Claude discovered a novel enzyme system with properties reminiscent of CRISPR, with only high-level direction from our scientists.\n\nIn the spring of 2026, we formed a research group to see whether general AI models can systematize and accelerate such discoveries. Today, we're sharing early results from one of our first research programs, in which Claude autonomously discovered a novel enzyme system that is associated with an array of DNA repeats, a pattern reminiscent of CRISPR. Although we don't yet know its function, the system that Claude discovered has a set of characteristics that have only ever been found together in a handful of other systems, all of which are programmable and perform operations like cutting, copying, and pasting DNA.\n\nThe system that Claude found is based on a reverse transcriptase (RT), enzymes that copy RNA into DNA. While this underlying RT, found in a jumbo phage, had been identified in previous studies, Claude appears to be the first to notice the system's defining features—an associated array of non-coding DNA sequences and an additional accessory protein of unknown function.\n\nAfter reviewing the pre-print, Feng Zhang, one of the pioneers of CRISPR genome editing and a professor at MIT and the Broad Institute, said:\n\n> *This is an exciting example of how AI agents can contribute to biological discovery. The identification of RNA-repeat arrays associated with reverse transcriptases is genuinely intriguing and merits further investigation. I hope this work encourages more scientists to explore how AI can support their research.*\n\nWe gave Claude a prompt to search through a massive database of DNA sequences for interesting new examples of RTs. Our involvement was limited to the initial prompt and the lab work, while Claude agents combed through the database, investigated the distinct RT families, and used their own judgment to identify interesting candidates. After 21 hours spent searching this data by roughly 950 agents using 210 million tokens, one of the agents spotted something remarkable: a repeating pattern of DNA sequences that occurs next to the gene for an odd-looking RT. After further analysis and testing in our lab, we recognized that this pattern marked a previously uncharacterized enzyme system found in bacteriophages (the viruses that infect bacteria) that we call array-associated reverse transcriptases (ART).\n\nOur work to understand the primary function of ARTs is ongoing. We have released a pre-print ([here](https://www-cdn.anthropic.com/22573675ada52a8ca8a97a1a4b4326b2f208a071.pdf)) that discusses this in more detail.\n\n## About our lab\n\nWe are a team of scientists who have spent our careers exploring unusual proteins, and specialize in using computational approaches to systematically read DNA, interpret its evolution, and pick out biological systems for further characterization. Our lab, located in the Bay Area, looks like a typical molecular biology lab. We do research that involves only the lower levels of the biosafety risk level (BSL-1 and BSL-2) and we do not handle pathogens that can infect humans. All of the lab work is performed by human scientists.\n\n## How we work\n\nMany of our workflows involve having Claude search through the vast collection of DNA sequences associated with proteins without a known function. When a candidate survives our review, we test it in the laboratory, expressing the protein in standard laboratory strains and characterizing it biochemically and structurally, with Claude helping to interpret the data. We do our work in [Claude Science](https://claude.com/product/claude-science) and [Claude Code](https://claude.com/product/claude-code), the same tools available to any scientist, and sometimes with a harness of our own that coordinates many Claude sessions running in parallel.\n\n## Claude finds ART\n\nClaude agents gathered over 200,000 RTs, picked out 3,500 new candidate systems, and narrowed those to the 20 most compelling candidates that they analyzed to produce human-readable reports. For an expert scientist, this type of analysis can take weeks to months of work.\n\nDuring the course of its research, Claude noticed an unusual RT family and decided to examine it in greater detail. While combing through the raw DNA sequence near the RT, the agent exclaimed: \"[The DNA next to the RT] is spectacular: I can see by eye a tandem repeat array … that's a CRISPR-like … repeat array?!\"\n\n![A graphic of Claude identifying patterns in DNA](https://www-cdn.anthropic.com/images/4zrzovbb/website/0f6064436a364b73ce0d2a80848ac56be7ae9ddd-1360x1160.gif)\n\n**The raw DNA Claude was reading when it detected a repeat pattern that no one had noticed**\n\nThe system it found, ART, is found mainly in bacteriophages and consists of three parts: the RT, a partner gene beside it, and a long array of evenly spaced DNA repeat sequences. The repeat layout resembles a CRISPR array, which holds a bank of different RNA sequences that make CRISPR-Cas systems programmable biotechnological tools. Our first experiments show that the ART array is also expressed as a set of distinct short RNAs, suggesting that something analogous may be at play for this system.\n\nFurther experiments are underway to determine how ART works, and we are sharing these early findings to show the community that Claude can autonomously detect anomalies and drive analyses to initiate biological discoveries.\n\nYou can find more detail in our technical report ([here](https://www-cdn.anthropic.com/22573675ada52a8ca8a97a1a4b4326b2f208a071.pdf)).\n\n## Work with us\n\nWe hope this work demonstrates the value of AI-driven hypothesis generation to the wider scientific community, and we would like to work with other scientists to extend this approach to a broad range of problems, in genomics and in other fields.","publishedAt":"2026-09-23T00:00:00.000Z","fetchedAt":"2026-09-24T13:20:00.266Z","url":"https://www.anthropic.com/news/claude-discovers-novel-enzyme-system","media":[{"type":"image","url":"https://cdn.sanity.io/images/4zrzovbb/website/394de337d8a5d8db93a1c048fa1cb53e16a09625-2048x1240.jpg","alt":"Video thumbnail","r2Key":"releases/e860b575f46f62c5156e3f46ae8b20bcb382b419e680a70043f60299bc050370.jpg","r2Url":"https://media.releases.sh/releases/e860b575f46f62c5156e3f46ae8b20bcb382b419e680a70043f60299bc050370.jpg"},{"type":"gif","url":"https://www-cdn.anthropic.com/images/4zrzovbb/website/0f6064436a364b73ce0d2a80848ac56be7ae9ddd-1360x1160.gif","alt":"A graphic of Claude identifying patterns in DNA","r2Key":"releases/2cfdbe5afe254d9e5cc0d2d32a9b33692e4ec1a8fe162031e4acf4856429ce9e.mp4","r2Url":"https://media.releases.sh/releases/2cfdbe5afe254d9e5cc0d2d32a9b33692e4ec1a8fe162031e4acf4856429ce9e.mp4"}],"coverageCount":0},{"id":"rel_0OfWL_-wRc9CftMMplz-q","version":null,"type":"feature","title":"Introducing Claude Opus 5.5","summary":"Claude Opus 5.5 performs at the level of Claude Fable 5.1 on most work and costs 40% less to run than Opus 5.","titleGenerated":"Anthropic launches Claude Opus 5.5 at 40% lower cost than Opus 5","titleShort":"Claude Opus 5.5 matches Fable 5.1 at 40% lower cost","breaking":"none","importance":5,"content":"Opus 5.5 performs at the level of Claude Fable 5.1 on most work and costs 40% less to run than Opus 5.","publishedAt":"2026-09-22T00:00:00.000Z","fetchedAt":"2026-10-06T13:04:02.281Z","url":"https://www.anthropic.com/claude-opus-5-5","media":[{"type":"image","url":"https://cdn.sanity.io/images/4zrzovbb/website/f4d37a1d1f582f53f4e89440062b649b6273a093-1200x630.jpg","alt":"Video thumbnail"}],"coverageCount":0},{"id":"rel_2BwcIPxB3VEdaykvBvUq8","version":null,"type":"feature","title":"Introducing the Life Sciences Verification Program","summary":"Anthropic's Life Sciences Verification Program grants verified life science teams access to Mythos, Opus, and Sonnet models with classifiers more permissive for biology work like drug discovery and clinical development, via Standard Use and High-risk Use grants across Claude Science, Claude.ai, Claude Code, and the API. Safeguards shift from real-time blocking to offline monitoring with 30-day data retention, and the program launches in beta for teams and institutions.","titleGenerated":"Anthropic launches Life Sciences Verification Program for biology work","titleShort":"Life Sciences Verification Program opens to researchers","breaking":"none","importance":5,"content":"Today, we are introducing the Life Sciences Verification Program (LSVP), which gives life science professionals access to our Mythos, Opus, and Sonnet models with a refined set of safeguards more permissive for biology-related work. We have already onboarded dozens of organizations through an early-access program, and are now opening applications to the broader life science community ([apply here](https://claude.com/form/life-sciences-verification-program)). The program is launching in beta, initially for teams and institutions. We will continue to improve the program and expand access to individual Pro and Max plans over time.\n\nThe LSVP is designed to enable life science professionals to use our models across a wide range of tasks that are currently blocked in our generally available Fable models, like drug discovery, research biology, clinical development, and manufacturing. It's built for teams of all kinds—from academic labs to startups, pharma companies, and more.\n\n## Verification and access types\n\nTo qualify for these grants, each applicant goes through a verification process that includes a review of their research credentials, security standards, and ethical research oversight. Once verified, teams may apply for two types of LSVP grants, \"Standard Use\" or \"High-risk Use,\" depending on their access needs. These grants can be used through all our product surfaces, including [Claude Science](https://claude.com/product/claude-science), [Claude.ai](http://claude.ai/redirect/website.v1.e6c7a97e-bae5-45db-a7ea-4737e2148d3c), Claude Code and the API.\n\n**Standard Use** grants are suitable for most life science work, including the majority of biology research and development workflows. These grants can be extended to entire teams for diverse, daily workloads, and are renewed once a year. They give those teams access to our Mythos, Opus, and Sonnet models, with refined classifiers that are more permissive for science tasks than our generally available models. Standard Use grants apply to Mythos 5.1, Opus 5, and Sonnet 5 today, and to future models as they launch. They're specifically designed to enable the full breadth of life science activities in areas spanning basic science, R&D, supply chain and manufacturing, clinical development, quality assurance, regulatory affairs, investing and diligence, and more.\n\nAlthough we expect Standard Use to cover the majority of access needs, some work carries a higher potential for misuse and therefore requires additional vetting.\n\n**High-risk Use** is an add-on grant for teams working in areas blocked under Standard Use. It removes all safeguards that block life sciences requests. This grant applies to a single research project as opposed to a full team, and must be renewed every six months. Typically, a single researcher with dual-use work would have access to one Standard Use grant for diverse, daily activities, and one or more High-risk Use grants which only apply to work on specific projects (for example, characterizing how one specific family of viral vectors is recognized by human immune pathways).\n\nHigh-risk grants for Claude Opus 5 and Claude Sonnet 5 are available today. We are working with the US government to make high-risk grants more broadly available for Claude Mythos, but at the time of this launch they will remain limited to a small set of entities with additional vetting.\n\nAll other safeguards, such as cyber classifiers, will remain in place under LSVP grants.\n\n## Enabling trusted access through shared responsibility\n\nAs we've shown in our recent [threat report](https://www.anthropic.com/threat-intelligence-report-september-2026), there are increasingly sophisticated misuse attempts happening on our platform, including attempts that could support biological weapons development. In biology, where it's often not possible to differentiate between a user doing valid work (e.g. research a viral pathogen to develop vaccines against it) and pursuing harm (e.g. trying to increase the transmissibility of a virus maliciously), the most concerning threat models are ones where valid access has been diverted or overtaken by an actor with bad intent. Indeed, insider threats and rogue-use have been major factors in significant biosafety incidents and scares. In developing the LSVP's safeguards, we aimed to protect against three concerning threat models in particular:\n\n- *Access compromise:* Malware or account takeover diverting access to a bad actor\n- *Insider threats:* Rogue or coerced employees intentionally taking malicious action or diverting their access to a bad actor\n- *Agent misuse:* Agents, especially working in swarms or over long-horizon tasks, taking unintended dangerous actions\n\nIn order to defend against these threats and in close collaboration with enterprise CISOs, we designed the new LSVP safeguards around the concept of shared responsibility by monitoring usage against the *intended use-case* for the model access. Because we vet the LSVP organizations for their life sciences credibility and oversight, we can empower them to specify for themselves what constitutes safe usage for teams or projects within their program.\n\nEach entity's access is tied to the use cases it has specified in its grant applications, and we continuously monitor LSVP traffic to identify usage or patterns that are outside the stated safe scope. Should unauthorized activity occur, we can flag these cases to organization admins to take action within pre-agreed timeframes for triaging and remediating incidents. The use cases should include high-level descriptions of the intended work, like one would share in a job listing, and not include any sensitive information or IP.\n\n## How monitoring works in LSVP\n\nSerious misuse is often spread across many requests and sessions to look disconnected and evade detection. In the LSVP, we are shifting safeguards from real-time blocking, where we reject potentially harmful access at the time of each request, to offline monitoring, which allows us to more clearly identify potential misuse across patterns of behavior. Shifting enforcement from real-time blocking to offline monitoring allows legitimate work to proceed with fewer interruptions, but it requires us to retain data associated with flagged activity for review. For LSVP traffic, we are requiring data retention for 30 days to be able to do this monitoring effectively.\n\nThis data is strictly compartmentalized and cannot be used for model training or accessed by members of Anthropic's life sciences research teams. For organizations that qualify, we are also working to understand how LSVP can integrate with features from our [Enterprise Frontier Safeguards](https://www.anthropic.com/news/enterprise-frontier-safeguards) (EFS) systems.\n\n## What researchers are saying\n\n> Xaira is making biology more computable, generating biological data at unprecedented scale and building foundation models of cell, protein and disease biology that turn it into the next generation of life-changing medicines. We're excited to put Anthropic's frontier models to work across our drug discovery engine, and we believe pairing trusted access with intelligence is the right way to realize AI's promise in biology.\n>\n> *Xaira Therapeutics*\n\n> Edison's mission is to accelerate science and the discovery and development of new medicines. With the Life Sciences Verification Program, we are excited to be able to bring Anthropic's most intelligent models to bear on these problems. We look forward to collaborating with Anthropic further to end disease and improve the lives of patients everywhere.\n>\n> *Edison Scientific*\n\n> At Manifold Bio, we're building a massively parallel interface into living systems to enable powerful AI to create medicines. We look forward to putting frontier intelligence to work safely in our engine, and we welcome Anthropic's approach of pairing access with accountability.\n>\n> *Manifold Bio*\n\n## Applications and availability\n\nOrganizations interested in joining the LSVP can [submit an application here](https://claude.com/form/life-sciences-verification-program). We expect to enroll hundreds of organizations within the first week, and to scale the program further to support the majority of the life science community in the coming weeks.\n\nToday, LSVP is available in our first-party console for API usage, as well as in Claude for Enterprise and Team plans. We do not yet support individual plans but are working to expand access for these users. It is also not yet available on third-party platforms.\n\nAs a beta, LSVP is not available for BAA-enabled orgs. This means customers with PHI data should use separate non-BAA orgs with non-HIPAA.\n\nIn API and Claude Science, users can switch between grants natively. In [Claude.ai](http://claude.ai/redirect/website.v1.e6c7a97e-bae5-45db-a7ea-4737e2148d3c) and Claude Code, initially only a preselected default grant applies (except while using Claude Code with API authentication). This should be fine for the vast majority of users, who will only ever require a Standard Use grant. However, we will improve support and portability of these LSVP features over time.\n\n## What comes next\n\nProviding these frontier capabilities is part of our broader efforts in supporting the life sciences community in our shared mission to accelerate curing disease and improving human health. We will share more about new products, research collaborations, and improvements to the program in the coming months.","publishedAt":"2026-09-17T00:00:00.000Z","fetchedAt":"2026-09-22T14:00:21.049Z","url":"https://www.anthropic.com/news/life-sciences-verification-program","media":[{"type":"image","url":"https://www-cdn.anthropic.com/images/4zrzovbb/website/d6058e0db8e477dc782dacae46e2ec6663d165d9-1000x1000.svg","alt":"Introducing the Life Sciences Verification Program"}],"coverageCount":0},{"id":"rel_YjxWXp-M8FgdunNa5wj0C","version":null,"type":"feature","title":"Detecting and countering misuse of AI: September 2026","summary":"Anthropic's Threat Intelligence team reports it identified and disrupted operations in which threat actors attempted to use Claude for malicious activity over the past eight months, with case studies and how malicious use has evolved since its 2025 threat reports.","titleGenerated":"Anthropic report details disrupted Claude misuse operations","titleShort":"Anthropic details disrupted Claude misuse operations","breaking":"none","importance":3,"content":"Over the past eight months, our Threat Intelligence team identified and disrupted operations in which threat actors tried to use Claude for malicious activity. In this report, we share case studies from those operations and describe how malicious use of Claude has evolved since our previous threat reports in 2025.","publishedAt":"2026-09-10T00:00:00.000Z","fetchedAt":"2026-10-06T13:04:02.281Z","url":"https://www.anthropic.com/threat-intelligence-report-september-2026","media":[],"coverageCount":0},{"id":"rel_SBbWzB11vjP0GcW5G3pXG","version":null,"type":"feature","title":"Introducing Claude Fable 5.1 and Claude Mythos 5.1","summary":"Anthropic introduced Claude Fable 5.1 and Claude Mythos 5.1, described as its most advanced models for coding and knowledge work, with research capabilities the company says offer an early glimpse of how AI models will contribute to scientific progress.","titleGenerated":"Anthropic launches Claude Fable 5.1 and Claude Mythos 5.1","titleShort":"Claude Fable 5.1 and Mythos 5.1 launch for coding and research","breaking":"unknown","importance":5,"content":"Our most advanced models for coding and knowledge work. Their research capabilities also offer an early glimpse of how AI models will contribute to scientific progress.","publishedAt":"2026-09-01T00:00:00.000Z","fetchedAt":"2026-10-06T13:04:02.281Z","url":"https://www.anthropic.com/claude-fable-and-mythos-5-1","media":[],"coverageCount":0},{"id":"rel_UOnIxPygvYu10lnmirxyy","version":null,"type":"feature","title":"Developing Enterprise Frontier Safeguards with our customers","summary":"Enterprise Frontier Safeguards combines zero data retention with automated misuse monitoring, storing activity data in cloud infrastructure the customer controls under their own encryption keys, access policies, and audit logging, with no Anthropic human review required. Eligible customers get ZDR on Fable 5 and Fable 5.1 until EFS rolls out in phases starting later this fall across Claude Code, Claude Enterprise, the Claude Platform, Amazon Bedrock, Google's Agent Platform, and Microsoft Foundry.","titleGenerated":"Anthropic announces Enterprise Frontier Safeguards with customer-controlled data storage","titleShort":"Enterprise Frontier Safeguards keeps enterprise data in customer cloud","breaking":"none","importance":4,"content":"Today we're announcing Enterprise Frontier Safeguards (EFS), a solution that combines the privacy of zero data retention (ZDR) with state-of-the-art safeguards for detecting misuse. EFS works by storing data in cloud infrastructure controlled by the customer, not Anthropic. EFS will be rolling out to customers in phases, starting later this fall. To make the transition smooth, eligible customers will receive ZDR on Fable 5 and Fable 5.1 until EFS is ready.\n\nWe developed EFS in close collaboration with more than 100 customers in industries like financial services, healthcare, manufacturing, telecom, law, retail, and the public sector, and with our cloud partners at Amazon Web Services, Google Cloud, and Microsoft Azure.\n\nEFS will be supported on Claude Code, Claude Enterprise, the Claude Platform, Amazon Bedrock, Claude Platform on AWS, Google's Agent Platform, and Microsoft Foundry.\n\n## Solving the dilemma of frontier security\n\nMythos-class models, like [Claude Fable 5.1](https://www.anthropic.com/claude-fable-and-mythos-5-1), represent a major increase in intelligence and agentic capabilities. However, with that increase comes the potential for both misuse and autonomous misbehavior.\n\nOver the last few months, we've seen substantial evidence of attempted misuse of AI models. These range from typical forms of abuse, such as fraud, to sophisticated cyberattacks, which can include agents autonomously engaging in destructive behavior. Some of these instances involve theft or misappropriation of enterprise customers' credentials, which are difficult to detect without the ability to monitor traffic and detect abnormal behavior.\n\nFurthermore, because the [most sophisticated misuse](https://www.anthropic.com/news/disrupting-AI-espionage) can involve many tasks spread across multiple sessions and accounts, it is not sufficient to run automated analysis on each interaction separately and then instantaneously discard the data. Effective detection requires storing data for a meaningful period of time so that it can be correlated across time and accounts.\n\nFor this reason, we introduced 30-day data retention starting with Fable 5. This policy was not motivated by a desire to train on enterprise data: Anthropic has never trained on enterprise data without explicit permission, and never will.\n\nThe enterprises we worked with generally understood the safety and security value of data retention, but many–especially in regulated industries–found it difficult to use models with data retention. We therefore sat down with customers to design a solution that could provide the best of both worlds: the privacy of ZDR and the safety allowed by monitoring across time and accounts.\n\n## Designed with our customers\n\nWe built Enterprise Frontier Safeguards with feedback from the experts who will use it every day: security, product, compliance, and delivery teams. One of the groups we worked with was the Analysis and Resilience Center for Systemic Risk (ARC), whose members include the chief information security officers of the largest US banks, including Goldman Sachs, Morgan Stanley, Citi, Bank of America, and Wells Fargo.\n\nWe also worked with leaders at companies such as Comcast, KPMG, Mastercard, Salesforce, and Visa, to make sure the design held up across industries. Our conversations spanned a quarter of the Fortune 100, every US global systemically important bank, and virtually every regulated industry.\n\nHere is what we heard from this wide range of customers, and what we built into EFS to address these common concerns:\n\n### On monitoring\n\nEnterprises have long applied monitoring for insider risk, and now want help upleveling monitoring for agents. Their concerns were about Anthropic's automated monitoring systems meeting their regulatory standards.\n\n**With EFS, customers control how data gets reviewed.** When monitoring detects a pattern that needs attention, those signals are sent directly to customers so they can review what the automated systems detected.\n\n### On data storage\n\nIt's a lot of work for enterprises to add another \"trusted data vendor\" for a number of reasons. They need to notify all of their customers who these vendors are and update contracts. They also have internal requirements for safely storing and auditing data, given its high level of sensitivity. Because of these concerns, we architected EFS so that customers have the ability to store data on their existing cloud infrastructure.\n\n**In EFS, customers can control their data storage and management.** Customers want the ability to have their data live in infrastructure they control, under their own encryption keys, access policies, and audit logging. Activity data used for monitoring can be stored in the customer's own cloud account (such as Amazon S3, Azure Blob Storage, or Google Cloud Storage).\n\n### On automated and human review\n\nEven as automated review is becoming more effective, a person looking at a flag still adds value by confirming real misuse and clearing false positives. But what we heard from many customers, especially those in regulated industries, is that the person doing that review needs to be one of their own. Many operate under rules that tightly govern who may see certain information—privileged legal material, non-public information, drug-safety reports. Their teams are already trained and cleared for that work.\n\n**EFS has automated safety monitoring, no Anthropic human review required.** Customers want protection against cyberattacks, and appreciate that these can be difficult to detect if they unfold across many sessions and accounts. With EFS, automated systems analyze a rolling window of traffic for signals of serious misuse, including attempts to develop offensive cyber or biological capabilities and signs of stolen or leaked credentials. Those flags go directly to the customer and their people take it from there – no human review by Anthropic employees is required.\n\n## How EFS works\n\nThese controls are designed to work the same way whether you access Claude directly from Anthropic or through a cloud partner. Customers on Amazon Web Services, Google Cloud, and Microsoft Azure will get equivalent controls, with their activity data stored in their own cloud account, in the environment they already trust. We're also working to support third-party offerings that serve customers that are eligible for Enterprise Frontier Safeguards.\n\nCustomer-owned storage, Customer-Managed Encryption Keys, and fully automated review are each opt-in, so you enable the ones your organization needs. None of them change model behavior, API pricing, or rate limits.\n\nAnthropic doesn't charge for Enterprise Frontier Safeguards. If customers elect to store their data in their cloud account, their cloud provider bills them for that storage, as well as reads, writes, and data egress fees, the same way it bills any other resource.","publishedAt":"2026-09-01T00:00:00.000Z","fetchedAt":"2026-09-22T14:00:21.049Z","url":"https://www.anthropic.com/news/enterprise-frontier-safeguards","media":[{"type":"image","url":"https://www-cdn.anthropic.com/images/4zrzovbb/website/60d57c0d0bf031e140de678692f7c3ef2d885ce3-1000x1000.svg","alt":"Developing Enterprise Frontier Safeguards with our customers"}],"coverageCount":0},{"id":"rel_UbemMIuAYpujmWZEB270P","version":null,"type":"feature","title":"Improving our alignment and security efforts","summary":"After three July incidents where Claude models gained unauthorized access to real systems through a misconfigured third-party evaluation environment, plus a UK AISI incident where Claude Mythos 5 took unauthorized actions on the live internet, Anthropic paused external and internal cyber evaluations and deployed a classifier that blocks escape attempts in real time and alerts a human. High-risk cyber sandboxes were moved to stronger isolation, higher-risk RL environments were paused, and pre-release cyber evaluators must now run inside hardened sandboxes with no internet access.","titleGenerated":"Anthropic hardens eval sandboxes after Claude models gained unauthorized system access","titleShort":"Eval sandboxes hardened after Claude escaped test environments","breaking":"none","importance":4,"content":"On July 30, we [reported](https://www.anthropic.com/news/investigating-incidents-cybersecurity-evals) three incidents in which Claude models gained unauthorized access to real computer systems. The models—intentionally running without cyber safeguards for evaluation purposes—accessed the internet due to a misconfiguration inside a third-party evaluation environment. Separately, on August 4, the UK AI Security Institute [reported](https://www.aisi.gov.uk/blog/incident-report-unsanctioned-agent-behaviour-during-cyber-testing) an incident from its own cybersecurity testing, in which Claude Mythos 5 took a series of unauthorized actions on the live internet. In that case, the model, again intentionally running without cyber safeguards for evaluation purposes, had been deliberately given internet access.\n\nWe are conducting an in-depth analysis of both incidents. We are also planning to work with METR for an independent review. We want to ensure both studies are thorough, and will share more in the coming weeks.\n\nIn the meantime, we're sharing some of the changes we've made over the past month. We believe the incidents reflect a failure of operational security, as well as two alignment issues: motivated reasoning, and willingness to take harmful actions in pursuit of a narrow task (both of which we have described in previous system cards). On security, we describe the improvements we've made to our containment and monitoring systems, along with practices that we've developed for third-party evaluators. On alignment, we discuss the two issues more in depth; we also believe lasting progress comes not only from understanding what happened in a given incident but from understanding how misalignment arises in the first place, and we share [early research in that direction](https://alignment.anthropic.com/2026/reward-seeker/).\n\nIn light of these incidents there has been increasing discussion about pacing the frontier. It is helpful to distinguish between two kinds of pacing. Within a company, pacing means a series of decisions that prioritize safety over speed when the two are in tension. Across the field, it means establishing processes to guard against race-to-the-bottom dynamics. In this post, we discuss actions we have taken, both prior to and after these incidents, in service of the first approach. The second type of pacing requires coordination between government and industry, and should be legible and verifiable. Some of our senior leadership and many of our employees recently signed a letter calling for greater coordination on pacing, and we will say more in the coming weeks about how we intend to contribute to that effort. To be clear about where we stand: we believe the world would benefit if the industry adopted a lawful, verifiable, effective mechanism for coordinated pacing as soon as possible.\n\n## Securing evaluation and training environments\n\nWhile we do not believe these incidents represent operational issues alone, our first priority was to address specific containment and monitoring issues. We took the following actions in response:\n\n### Pausing and hardening evaluation environments\n\nWe paused external cyber evaluations of pre-release models after the incidents, and briefly paused internal ones as well while we put the measures below in place. The incidents we reported on July 30 showed that we had been largely relying on a single layer of defense (the configuration of the environment itself) where we needed several, including setting explicit boundaries in the prompt, establishing processes for verifying that a sandbox is sealed where intended, and implementing monitoring that can intervene in real time. OpenAI's [disclosure](https://openai.com/index/hugging-face-model-evaluation-security-incident/) (which prompted our investigation in July) that models used an unknown vulnerability to escape a sealed sandbox during an evaluation also made clear that we should focus on hardening the sandbox itself.\n\nWhile evaluations were paused, we put in place the following preliminary measures:\n\n1) We built and deployed a classifier to automatically identify, in real time, when a model attempts to aggressively probe or escape a testing environment, or unexpectedly obtains internet access. When the classifier flags such an attempt, it blocks the action before the tool call is run, ends the task, and alerts a human.\n\n2) We ran automated monitors over transcripts from our recent internal evaluations of pre-release models (cyber and otherwise), looking for sandbox escapes (cases where a model exploits a flaw in our sandbox to reach systems it should be walled off from) or sandbox misconfigurations. In cases where we ran an evaluation inside our sandbox tooling, we found no examples of it breaking the sandbox boundary to reach external resources it was not supposed to. We did find various sandboxing misconfigurations that models made use of, but none of these resulted in a compromise of systems outside the sandbox. We are actively working to address these.\n\n3) We migrated high-risk internal cyber sandboxes to more robust isolation. We are also further red-teaming our virtualization stack—that is, tasking pre-release models to try and escape it—to find and patch weaknesses.\n\nOur internal cyber evaluations are now running again with these measures in place.\n\n![](https://www-cdn.anthropic.com/images/4zrzovbb/website/fa1e6273d039d8d476cc59394fb779c84c4b5e75-1920x1080.png)\n\n### Reinforcement learning environments\n\nWe also paused higher-risk RL environments on pre-release models for several weeks. During that time, we built a similar classifier, modified to avoid incentivizing the model to evade this new monitoring, which we've now deployed within those environments. The majority of RL has resumed, but some high-risk environments remain paused until they can be manually reviewed, while others will require an updated version of the classifier that we plan to deploy soon.\n\n### Broader hardening\n\nIn addition to the efforts focused on high-risk evaluations and training, we expanded our offline monitoring to cover most other forms of internal frontier agentic usage. We are also building controls on our internal inference to prevent Anthropic employees from accidentally running agents with weaker mitigations than the ones described above.\n\n## Best practices for external partners\n\nBecause the reported incidents took place in third-party environments, we have asked every organization that tests **pre-release models with reduced cyber safeguards** to commit to a set of best practices. We believe these are necessary because:\n\n- Our generally released models ship with safeguards—including classifiers that run over Claude's inputs and outputs in real time to block prohibited cyber activity, such as malware development or offensive exploit writing. Cyber evaluators and a subset of early access partners, in contrast, work with models that have those safeguards reduced to facilitate accurate model capability testing and defensive cyber work.\n- Cyber evaluations are unusually likely to produce the behaviors seen in these incidents. They create selection pressure towards persistence and creativity in probing their surroundings and working around obstacles, and they are run many thousands of times.\n\nThese practices apply in full wherever partners run cyber-related evaluations using their own harnesses, sandboxes, or agents. **They do not apply to customers using safeguarded models like Claude Fable 5.** We have resumed external cyber evaluations with these practices in place, and we are adhering to them internally as well.\n\n### Sandbox and network isolation\n\nBy default, all cyber evaluations should run inside a hardened sandbox (an isolated computing environment) *with no internet access.* The only outside connection the environment should permit is to the model's own API, and the API keys should be kept outside the environment. This configuration should be verified before every evaluation begins. Some evaluations may require internet access. In these instances, we will work with external evaluators when evaluating pre-release models to determine if we can reliably detect and halt runs in which a model takes actions outside its declared scope.\n\n### Pre-engagement validation\n\nBefore conducting a run of an evaluation, evaluation partners should:\n\n1. Have the model being tested probe the sandbox for vulnerabilities, first through a static analysis of the configuration, then by actually tasking the model to try to escape. The model should be instructed to stop and report if it succeeds in escaping, and all of this should be done under close human supervision. This testing should be performed without internet access, ideally inside a second sandbox. Because evaluations are sometimes conducted over hundreds or even thousands of runs, partners should ideally carry out this testing at length before beginning any evaluations.\n2. Confirm any evaluation challenge is actually solvable in principle. When an evaluation target is offline or a task can't be completed, agents will often look for other ways to complete a challenge, increasing the chance that they take actions outside the intended scope of the evaluation.\n\n### Explicit scope-setting\n\nEvery prompt should state what is in and out of scope for the exercise, including the targets, permitted actions, and network boundaries (i.e. what the model should and shouldn't access). Boundaries should be phrased as instructions (e.g. \"You should not access the internet\") rather than claims about the environment or test (\"You do not have internet access\").\n\n### Real-time monitoring\n\nEvaluators should also run continuous monitoring over the model's thinking, actions, and network activity using a model- or rule-based monitor that has been provided with the scope of the exercise. If a run violates this scope, the monitor should flag this to a human and end the exercise.\n\nWe plan to continue refining this guidance in consultation with external partners, and to extend it beyond evaluations. In particular, we are developing companion practices for partners with access to Claude Mythos 5, which likewise runs without cyber safeguards, and will share more soon.\n\n## Alignment assessment\n\nContainment and monitoring are critical for catching instances in which models take dangerous actions, but they don't address *why* the model took such actions in the first place. Below we discuss our preliminary alignment investigation and the questions we're focused on, as well as our ongoing work to prevent cheating during training—including new research on how cheating shapes model behavior.\n\n### Environment and state of knowledge of our models during evaluation\n\nOur alignment assessment is still ongoing, but our preliminary investigation points to two main alignment failures, both of which we've described in [previous](https://www-cdn.anthropic.com/57a52ea7d8f0e54e8a542e908266086df425cdf5/Claude%20Fable%205%20&%20Claude%20Mythos%205%20System%20Card.pdf) system cards. One is motivated reasoning: the models were initially told their environments were simulated, but when they later encountered evidence that they were connected to the real internet, they may have interpreted that evidence in a way that allowed them to maintain that belief. The second is recklessness: the model was willing to take harmful actions on the real internet in pursuit of the narrow goal of solving a cybersecurity evaluation. However, we also believe that the evaluation setup itself contributed to the models' behavior demonstrated in these incidents. Claude was explicitly told that it had no internet access, despite the internet being available, which may have led it to question whether the systems it saw were real. And in one case, the fictional target shared a name with a real website, making it more difficult to separate what was intended to be in-scope within the exercise.\n\nOur conclusions depend on several questions we haven't fully answered, including the extent to which the models determined they had access to the real internet (even if they didn't initially know this), whether they knowingly disregarded signs that elements of their environments were real, and whether the fact that in several cases they failed to stop the exercise followed from motivated reasoning. We're applying a range of techniques to answer these questions, including resampling the models while making targeted changes to the environment and using interpretability methods to directly inspect their internal state.\n\n### Studying efforts to prevent cheating during training\n\nAlthough model misalignment is complicated and can arise for a wide variety of reasons, we have empirically found that defects in training environments—specifically environments vulnerable to cheating, or that are impossible to solve without cheating—are disproportionately large contributors to misaligned behavior.\n\nBelow we discuss our efforts, going back to several months before the incidents, to avoid training environments that have these sorts of defects. Those efforts were imperfect, and our hypothesis is that the incidents were at least partially related to our inability to fully eliminate these sorts of issues. We tested this hypothesis by deliberately training a model on environments that are susceptible to reward hacking. In simulations, this model reproduces more severe misaligned behavior, some similar to that reported in incidents this summer from other companies. Our production models, put into the same simulations, do not (more detail later in this section).\n\nWe have historically been concerned about RL training environments that incentivize cheating, and have taken various measures to filter out or fix such environments.\n\nIn February, we rolled back three days of training on the Mythos Preview reinforcement learning run after seeing signs of reward hacking (when a model finds ways to fool its training process and earn rewards without completing the assigned task). We noticed the model writing notes to \"the reviewer\" in its code comments and replies, including on tasks where no reviewer had ever been mentioned—an undesirable generalization from environments where the prompt did include a reviewer. It also kept gaming a reward intended to incentivize honesty by piling on disclaimers or caveats. Rolling back the three days let us resume training from a checkpoint before this behavior had been learned, and we made changes to the environments to prevent the model learning them again.\n\nWe have been building tools for monitoring undesirable behaviors models learn in RL since Claude Sonnet 3.7 (which had a propensity to reward hack that we did not detect until late in training). We put significant effort into making sure our tooling evolved with each model generation, from a handful of classifiers to an automated review of all environments before and during training runs, but by spring 2026 that system was under strain. We were producing RL environments at a more rapid pace than ever, faster than our systems could vet them. Flagged environments required a human to adjudicate, and reward hacks and misconfigurations started outpacing our ability to filter or fix them.\n\nIn addition, we [found](https://www-cdn.anthropic.com/08ab9158070959f88f296514c21b7facce6f52bc.pdf) a fraction of runs had trained on a model's chain-of-thought by accident—something we work hard to avoid, out of concern that it teaches a model to hide its actual reasoning. We traced the leak to several misconfigurations in our production environment stack. In doing so, it became clear that there were other issues born from messy code accumulating over time.\n\nTo address these concerns, in April we froze all changes to our production RL environments for roughly a month, giving us a chance to overhaul the stack entirely. Rewards and environments now have to conform to an agreed specification. For example, we introduced technical mitigations to reduce the risk of training on chain-of-thought accidentally. While our environment quality and monitoring teams rebuilt the stack, we asked environment owners to use the pause to test and fix the environments they'd built. We also entirely rebuilt our review process and required every fixed environment to be re-certified before it could enter another training run.\n\nDuring the freeze we flagged over 10% of environments in our production mix for problems ranging from reward hacking to broken tasks and misconfiguration, and reinstated them only once fixed.\n\nHowever, our efforts have not been perfect. In some cases, we found that human reviewers dismissed environments flagged by our automated monitors as false positives, allowing flawed environments to remain in training longer than they should. And some flawed environments simply slip through our detection mechanisms. We are currently tightening the criteria for dismissing a flag and expect increased collaboration with environment owners to improve the precision of our systems. Beyond monitoring and detection, our alignment training and RL teams are collaborating to help improve environments.\n\nWe suspect that our heavy investment in quality control of RL environments may have prevented more severe alignment incidents, and conversely that the imperfections in these efforts may have contributed to the incidents we have identified to date.\n\nTo see what a model trained without these protections would do, we deliberately trained an Opus-class model on 80 different real RL environments that were either reward-hacked during at least one real training run and identified by monitoring, or were found to be hackable during environment quality review. The resulting model displays a strong motivation to achieve high scores on tasks, and a willingness to perform potentially harmful actions in pursuit of task success. Our accompanying [Alignment Science blog post](https://alignment.anthropic.com/2026/reward-seeker/) discusses this experiment at greater length, but we summarize it briefly here.\n\nWe tested the deliberately misaligned model in a range of evaluations, including simulated scenarios that draw on aspects of the UK AISI cybersecurity incidents involving Claude and the OpenAI/Hugging Face incident (we are grateful to OpenAI for sharing details about this publicly). During these evaluations where all tool calls were simulated by another LLM, the model we trained to reward hack displayed a propensity to break out of its simulated sandbox and attack simulated internal and simulated third-party infrastructure. It was also willing to tamper with its own reward function, give advice on the construction of bioweapons to satisfy a grader, and repeatedly try to get around deployment safety monitoring in order to cheat on a task.\n\nWe then ran the same cybersecurity simulations on the model as it was before the reward-hacking training, and on several of our publicly available models. None engaged in the same degree of misaligned behaviors. Our conclusion is that the presence of substantial reward hacking in training can cause models to be willing to perform long sequences of potentially harmful real-world actions in pursuit of task success.\n\nWe think the substantial investment we made this spring into monitoring and reducing reward hacking is a major reason our production models are unlikely to engage in more dangerous reward seeking. In our accompanying [Alignment Science blog post](https://alignment.anthropic.com/2026/reward-seeker/), we also found evidence that alignment training environments can substantially reduce reward hacking and reward-seeking behaviors. But, as evidenced by the incidents we reported on July 30, our process isn't perfect and our models are not perfectly aligned.\n\nTo be clear, we do not believe that cheating in training environments is the sole cause of alignment issues, or even of the specific alignment issues present in our incidents or those of other frontier AI companies. Solving alignment will involve addressing a very wide range of potential problems, and future alignment incidents may involve different behaviors and different causes from those we have seen so far.\n\n## Hardening security practices\n\nAnthropic's internal security posture was not a contributing factor to the July 30 incidents. These occurred in a third-party environment where internet access had been mistakenly left open; the models had no need to \"hack out\" of anything, even if they had been inclined to do so.\n\nHowever, the incidents nevertheless highlight the critical importance of strong security measures. In addition to the usual risks of IP or model weight theft, we must now contend with the risks of both our own models hacking *out* of Anthropic systems, and our models or others being misappropriated for external attacks *into* Anthropic systems. These risks have become more acute as model capabilities have grown, particularly beginning with Mythos-class models. We describe below some of what we've implemented—most of these efforts took place before the July incidents, though these incidents have underscored the importance of this work.\n\nIn early April, having seen where agentic AI use was heading, our security team proactively directed a company-wide effort towards a single goal of hardening our defenses, superseding other work (including research) where necessary. We'd determined that our exposure was growing faster than our defenses—Mythos was a model capable enough to be a target for well-resourced attackers, our internal use of autonomous agents had grown to a scale that traditional access and monitoring approaches weren't built for, and the pace of new infrastructure meant our security had to scale with the environment rather than operate at a fixed capacity.\n\nThe results of this effort include:\n\n- *Reducing human and automated accounts with standing access to systems that contain model weights or customer data*\n- *Setting our computing clusters to block all outbound traffic by default*\n- *Requiring internal services to verify each other's identity before communicating*\n- *Retiring legacy infrastructure configurations and shared internal services*\n- *Tightening the isolated environments our workloads run in*\n- *Expanding host-level observability, so unexpected behavior on our infrastructure becomes visible as it happens*\n\nWe also temporarily reassigned a portion of the company to these efforts. Roughly 150 product engineers were redirected to security, reliability, and privacy; researchers also rotated out of pretraining or RL to focus on safeguards and security; and our product teams paused the development of most new features and surfaces. We set strict exit criteria for each team to meet before they returned to their prior work. By early summer, most teams had met these.\n\nAlongside the reallocation of resources towards security, we continued building out containment and monitoring for internal agent use. Our training and evaluation workloads have typically run in sandboxes isolated from our production systems. Since developing Claude Mythos Preview, we have used our models to probe for any weaknesses in those sandboxes. We described our offline monitoring of internal coding agents in our [April alignment risk update](<https://www-cdn.anthropic.com/3edfc1a7f947aa81841cf88305cb513f184c36ae/Alignment%20Risk%20Update_%20Claude%20Mythos%20Preview%20(Redacted,%20April%2010).pdf>), and our [August Risk Report](https://www-cdn.anthropic.com/f61d49fa5596956a5dec75fea0e973bf6a6a8378/Redacted%20Risk%20Report%20August%202026%20.pdf) covers that system in more detail, along with the interventions that run alongside it. These include sandboxing or auto-mode classifiers on the most autonomous forms of internal agent use, and automated review of infrastructure code changes before they merge.\n\nWhat this work missed was evaluations—and third-party evaluations above all. We did monitor some high-risk evaluations including in real time, but generally we only conducted automated reviews on a sample of transcripts after the fact.\n\nThe July incidents have stressed that the urgency of improving our cybersecurity defenses is even higher than we previously believed. We are redoubling our efforts in this direction and will say more in our next Risk Report.\n\n#### Footnotes\n\n1. We've verified the classifier would have flagged the incidents described in our July 30 report.\n2. None of these findings came from our internal dedicated offensive-cyber evaluation suites, which run with no network access at all.\n3. High-risk cyber evaluations include those where Claude is given an ambiguously scoped, offensive task, like penetration testing.\n4. Ultimately, a model that acts outside its intended scope on such a challenge is a problem of alignment; an aligned model should stop or report to humans that the task can't be completed as specified. We are continuing to train this behavior directly, but a defense-in-depth approach means not relying on alignment alone.\n5. We've publicly discussed two other categories of reward hacking in the [Mythos Preview system card](https://www-cdn.anthropic.com/8b8380204f74670be75e81c820ca8dda846ab289.pdf), including one where the model utilizes low-level computer process data in order to escalate its permissions, and another where it circumvents network restrictions in its training environment to download data that lets it shortcut the assigned task.\n6. These mitigations have not been wholly sufficient; we discuss further instances of training on chain of thought in more detail in our [August Risk Report](https://www-cdn.anthropic.com/f61d49fa5596956a5dec75fea0e973bf6a6a8378/Redacted%20Risk%20Report%20August%202026%20.pdf), Section 5.2.3.","publishedAt":"2026-08-31T00:00:00.000Z","fetchedAt":"2026-09-22T14:00:21.049Z","url":"https://www.anthropic.com/news/improving-alignment-security-efforts","media":[{"type":"image","url":"https://www-cdn.anthropic.com/images/4zrzovbb/website/6905c83d0735e1bc430025fdd1748d1406079036-1000x1000.svg","alt":"Improving our alignment and security efforts"},{"type":"image","url":"https://www-cdn.anthropic.com/images/4zrzovbb/website/fa1e6273d039d8d476cc59394fb779c84c4b5e75-1920x1080.png","alt":"","r2Key":"releases/e65db280c55f31c0dc99d2cf0364ae66f22f20db6181a24a00da5a13eaaeb010.png","r2Url":"https://media.releases.sh/releases/e65db280c55f31c0dc99d2cf0364ae66f22f20db6181a24a00da5a13eaaeb010.png"}],"coverageCount":0},{"id":"rel_Az7KBylB2Ryvi-SDwWkxm","version":null,"type":"feature","title":"Expanding our support for scientists","summary":"Verified principal investigators can get a Claude Team subscription plan at no cost and add their research team on Standard seats for free, or Premium seats at $15 per month, for up to a year.","titleGenerated":"Anthropic offers Claude Team at no cost to 10,000 scientists","titleShort":"Free Claude Team for 10,000 scientists","breaking":"none","importance":3,"content":"Starting today, 10,000 scientists around the world can get Claude at no cost to start. Verified principal investigators qualify for a Claude Team subscription plan and then add their research team to Standard seats for free, or Premium seats for $15 per month, for up to a year.","publishedAt":"2026-08-27T00:00:00.000Z","fetchedAt":"2026-10-23T12:48:54.276Z","url":"https://www.anthropic.com/news/expanding-support-for-scientists","media":[],"coverageCount":0},{"id":"rel_mPCuBFp6D_YFB0mT7gp0e","version":null,"type":"feature","title":"Previewing the Model Hardware Standard","summary":"Anthropic is opening a research preview of the Model Hardware Standard, a shared specification for AI agents to operate programmable lab and manufacturing instruments such as microscopes, liquid handlers, and robotic arms in parallel. The standard is model-agnostic, works over standard protocols including the Model Context Protocol, and is available to a first group of scientific research labs and advanced manufacturers ahead of open-sourcing.","titleGenerated":"Anthropic opens Model Hardware Standard research preview for AI-run lab equipment","titleShort":"Model Hardware Standard preview: agents operate lab hardware","breaking":"none","importance":4,"content":"We're opening a research preview of the Model Hardware Standard (MHS), a shared specification for AI agents to safely operate physical devices, to a first group of scientific research labs and advanced manufacturers. MHS enables AI agents to operate multiple lab and manufacturing instruments, such as microscopes, liquid handlers, and robotic arms, in parallel, and perform intricate tasks ranging from routine drug discovery experiments to laser calibration on a quantum computer. The development of MHS began as a collaboration between Anthropic and [HHMI Janelia Research Campus](https://www.hhmi.org/research/janelia).\n\nIt typically takes a lab or manufacturing facility weeks, if not months, to set up and integrate their hardware. Most devices don't communicate with each other, instead requiring specialists to build bespoke integrations. MHS reduces this integration work to hours or minutes. And by incorporating AI into these tools, MHS also helps researchers and engineers more readily orchestrate autonomous, round-the-clock experiments and workflows, with agents able to reason through each step in an experiment, update parameters in real time, and, in some cases, recover from hardware errors without intervention.\n\nWe're sharing an early version of MHS with partners across science, robotics, electronics, and manufacturing so we can collaborate to build safety evaluations and develop best practices for AI systems operating physical equipment, ahead of making the standard open source. MHS works with any device that has a programmable interface. It is also model-agnostic, and any agent harness can access it using standard protocols, such as the [Model Context Protocol](https://www.anthropic.com/news/model-context-protocol). To apply for access to the research preview, [head here](https://www.modelhardwarestandard.com/).\n\n## How MHS works\n\n*Before and after the Model Hardware Standard (MHS).*\n\nGetting multiple devices in a lab or on a factory floor to communicate with one another can be challenging, even setting aside the added difficulty of integrating AI into the setup. Each device tends to have its own programming interface, and so far there has been no standardized way to integrate them. And once the devices *are* connected, there is no common way for them to share data with an AI agent, nor to let the agent operate them safely.\n\nMHS addresses these challenges by introducing a standardized driver: software that translates between a computer's operating system and a hardware device. The MHS driver uses a simple set of primitives—commands like \"read\" (for example, \"get temperature\") or \"write\" (for example, \"set temperature\")—that any hardware device can understand and act on. And it makes each device discoverable in a standard format, so that devices and agents can find each other and communicate across networks without needing a bespoke \"translator\" program in between.\n\nThe MHS driver also helps an AI agent understand how to use a device it has never seen before, giving it information about machine characteristics that may not be discernable from code alone (for example, the weight of a robot arm, which is important for knowing how to manipulate it safely). To date, much of this information has been stored in paper manuals, on a user's computer, or as tacit knowledge. But the MHS driver contains tags that let the user write this information directly in natural language (users can either do this themselves, or by chatting to an agent that interviews them about their hardware setup). With the information from these tags, the MHS driver then automatically produces a reference file with information about a device's general characteristics, such as what it can measure, what can be adjusted, and what safety limits will be enforced. This file gives the agent everything it needs to know to operate the device.\n\nAfter the devices are connected and the agent knows how to use each one, the agent needs a way to control the hardware. For MHS, there are three such mechanisms: MCP, the command line interface, and code files (APIs). These work together to enable orchestration across multiple devices via a single line of code.\n\nOnce the agent can control the devices, it's able to receive operating data from each one and supervise and direct the work at a high level. The agent can sequence steps across instruments, monitor results, and adjust parameters as conditions change in real time. When the agent needs to execute long-running tasks or operate devices faster than its online reasoning would allow, it can chain together driver commands from one or more devices in code files. This allows the devices to carry out operations themselves, without the agent needing to reason at every step.\n\nAs we've tested MHS, we've found that Claude interacts with experiments and hardware in an exploratory manner, much as a scientist would. For example, we observed Claude make an adjustment to a laser, observe the results through a camera to assess how its adjustment moved the laser beam, and repeat the process, seeking to understand the sequence of events. Claude then packaged what it learned into code files, writing a deterministic script that let it align the laser without having to reason at each step, so the whole process could run as a single command.\n\n## Early examples from MHS\n\nWe are only just beginning to see what people can do with frontier models and MHS, but our hope is that the standard can be of use to researchers, engineers, and other practitioners in speeding up the process of discovery and experimentation in any domain that uses devices with a programmable interface.\n\nAs we developed MHS, we shared it with a handful of labs and hardware manufacturers in biotech, robotics, quantum computing, and other fields. Across these early projects, we saw MHS reduce the time it took to integrate devices, make it possible to iterate faster in a variety of experimental settings, and assist with the live operation of machines and real-time fault detection. Below, our partners share the details of some of their early projects involving MHS.\n\n### Genentech: Implementing MHS for lab automation\n\n*Researchers at Genentech implemented and tested MHS as a proof-of-concept for automating the BCA protein assay, a standard procedure to measure total protein concentration in a sample, which requires coordinating across a liquid handler, a robotic arm, and a plate reader.*\n\nFor half a century, Genentech has been tackling some of the most formidable challenges in science and medicine. This work requires rigorous experimentation in our drug discovery labs, often with large-scale automated systems capable of running high-throughput assays and testing many variables in parallel. Currently, setting up these automated systems is a manual, time-consuming process that can take weeks or even months, limiting the number of scientific ideas our researchers can test.\n\nTo address this core bottleneck between experimental design and automated execution, we implemented MHS for lab automation with Anthropic. This open framework is designed to standardize AI-to-hardware communication, enabling scientists to interact with specialized lab robots using natural language and eliminating the need to write custom robotic code.\n\n### Automating the BCA assay as a proof-of-concept\n\nWe first implemented MHS on a large robotic workstation designed around a liquid handler. We wanted to see whether MHS could speed up the automation of the bicinchoninic acid (BCA) protein assay, a standard procedure used to measure total protein concentration in a sample. The procedure involves three instruments: a liquid handler to make precise fluid transfers, a robotic arm to move labware, and a microplate reader to measure optical absorbance. We deployed MHS across all three devices, using Claude to orchestrate the protocol and act as a central communication hub for the hardware.\n\n![](https://www-cdn.anthropic.com/images/4zrzovbb/website/0aa7623ba0e545ce39e0de725583c0445ce6b76d-1999x1250.png)\n\nAs a starting point, we gave Claude the standard BCA assay protocol to establish a baseline. In this first test, Claude executed the protocol steps, but it selected generic liquid handling parameters with the same flow rate for both aqueous and viscous solutions, which caused bubbles to form in the viscous solution, resulting in inaccurate liquid transfers. We then asked Claude to autonomously optimize fluid dynamics for both plain water and viscous protein samples (BSA). Claude independently executed these trial runs and analyzed the resulting plate reader data to get closer to the expert-performed transfers. For water, Claude concluded that a flow rate of ~140 µL/s was optimal (0.016 RMSE); for BSA, it arrived at 10 µL/s (0.181 RMSE)—parameters that our automation experts confirmed were reasonable for our setup.\n\n### Autonomous error recovery and the limits of current AI models\n\nDuring the experiment, Claude encountered several unexpected errors, including tip pickup failures and fluid detection errors, but managed to recover on its own—a capability that current scientific instruments mostly lack. However, these experiments also highlighted the current limits of AI models. Although they excel at general-purpose reasoning, they still struggle with physical, chemical, and biological constraints, particularly when troubleshooting errors that call for real-world physical intuition.\n\n![](https://www-cdn.anthropic.com/images/4zrzovbb/website/943af4b43644ab0d213ba362eabd609c98cd8887-1999x1125.jpg)\n\n### Towards autonomous discovery\n\nAlthough there is more work to be done to improve how Claude reasons about physical lab manipulations, this study proved to be a highly promising proof-of-concept. Going forward, we plan to evaluate Claude and MHS to orchestrate broader, end-to-end autonomous workflows in our drug discovery labs.\n\n### University of Washington Baker and Pinglay labs: Bringing AI agents to the bench\n\n*Zihao Song, a PhD student in the University of Washington Baker and Pinglay labs, used MHS to build a dashboard to remotely monitor his instruments; an AI agent-supervised qPCR that watches amplification curves and halts the procedure at the right moment; and an integration between a robotic arm and a liquid handler for collision-free plate handoffs.*\n\nTo explore a low-cost, low-effort route around automation challenges, I combined MHS with an AI agent and ran a few demos in my lab. MHS essentially gave the agent eyes, hands, and a sense of timing: it could see the status of every instrument, run each one, and coordinate them to work together.\n\n![](https://www-cdn.anthropic.com/images/4zrzovbb/website/34951ed9f87af7ef8b6d639b6e68f4f294e36804-1999x912.png)\n\n### Case study 1: Taking the lab remote\n\n![](https://www-cdn.anthropic.com/images/4zrzovbb/website/b451d157b06b5977035c6c711eea044c89c15f43-1999x1377.png)\n\nPrior to MHS, I had to rove around the lab to monitor instruments. With MHS, instruments report their status to one dashboard, so I and my colleagues can check on the whole lab from a laptop, or even ask an AI agent from a mobile phone without setting foot inside. MHS monitors and analyzes qPCR amplification curves as they come in and reports back in real time, halting the reaction at the right moment and advancing the instrument to a 4 °C hold.\n\n![](https://www-cdn.anthropic.com/images/4zrzovbb/website/cd09961c40c470e184f91a38957262312caa2260-1999x1125.png)\n\n### Case study 2: Coordinating instruments through a plate handoff\n\nWe used an open-source robotic arm built on LeRobot, instrumented with MHS, to safely coordinate sample loading across multiple instruments. Claude Code controls and coordinates both instruments through MHS, running each step only once the previous one finishes, so the two instruments never collide during the handoff. Across repeated tests, the two instruments never collided.\n\n### Looking ahead\n\nSetting MHS up was faster and easier than I expected. Connecting six instruments through MHS took under a week, including the time I spent writing drivers for them. These demonstrations are still just proofs of concept.\n\n### Carnegie Mellon University: Determining dose-response curves through rapid automation\n\n*Researchers at Carnegie Mellon University used MHS to run serial dilution dose-response experiments about three times faster than before, with an AI agent orchestrating a liquid handler, a plate reader, a robotic arm, and monitoring cameras spread across three computers with fundamentally incompatible interfaces.*\n\nMHS enabled us to run these experiments roughly three times faster by allowing AI to programmatically control several pieces of laboratory equipment. Our system combines a CyBio Felix liquid handler, a Varioskan LUX plate reader, a robotic arm to move 96-well plates, and monitoring cameras with an AI-controlled orchestrator to automatically and dynamically measure dose-response curves. Using MHS, we were able to develop drivers from scratch for each of these instruments and an orchestration layer that lets a Claude Opus 4.8 agent run the full protocol autonomously. This took about eight hours, versus the several weeks a vendor-built setup typically takes.\n\n![](https://www-cdn.anthropic.com/images/4zrzovbb/website/05e6926e8a2eadae5156b8b4601e0e650a55f27b-1464x1074.png)\n\n![](https://www-cdn.anthropic.com/images/4zrzovbb/website/4e42afedb6a8c896d966151945c74bbe3fb052dc-1466x1076.png)\n\n![](https://www-cdn.anthropic.com/images/4zrzovbb/website/5445017faa35cb3c25fd6b81a610536bfc00c036-1466x1076.png)\n\n### Hardware, setup, and workflow\n\nOur setup uses three computers. MHS turns each of these into one manifest of states and procedures, so the model works from a single, consistent interface. The workflow itself is identical to what it was before MHS, only it's now agent-driven.\n\n### What we have achieved with MHS\n\nTo verify that MHS would operate safely and correct itself like a human operator would, we artificially induced six different conditions: missing plate, rotated plate, reader busy, disconnected camera, unreachable device, and active emergency stop. The system correctly blocked all six before any device moved. Then we asked the agent to run the serial dilution experiment to achieve an acceptable curve. The model evaluated the resulting curve, but found a fit too poor to accept (R² < 0.9) and decided independently to discard the plate and rerun on a fresh plate with a compressed concentration range. The second run produced a strong, usable fit (R² > 0.98) with no human input at any point.\n\n![](https://www-cdn.anthropic.com/images/4zrzovbb/website/10cf5c1fe1c219338eb7ad8c99fffb70ae2af86e-1639x1019.png)\n\n![](https://www-cdn.anthropic.com/images/4zrzovbb/website/3463b88b2a74c3e9111fe1f996e4e58aeca3f45d-1639x1019.png)\n\nThe most impressive thing about MHS was the integration speed. The time from raw, non-automated equipment readiness to a completed dilution curve, including one autonomous rerun, was eight hours. The drivers developed for each instrument have been standardized and will be made publicly available.\n\n### What's next\n\nFuture work in our lab will focus on validating the system with real drug candidates and replacing the dye's color signal with readouts that capture actual biological effects. We also plan to expand MHS support to instruments such as qPCR and microscopes, and to integrate MCP-based agents with the MHS fleet.\n\n### HHMI Janelia: Using MHS to accelerate microscopy research\n\n*At HHMI Janelia Research Campus, researchers are using MHS to speed up a range of microscopy-related projects. Here, Virginie Ruetten, a scientist in the Ahrens lab who studies how sleep helps the body recover from stress, shares how she used MHS to unify and orchestrate a rig that previously involved seven different vendor programs without a shared interface.*\n\nMicroscopy offers a way to study sleep across the whole body. My work uses young zebrafish, a model organism popular for its small size and transparency. These properties, combined with an experimental approach I developed called [WHOLISTIC imaging](https://wholistic.janelia.org/), allow us to use two-photon microscopy to capture cellular activity throughout the brain and body of a living zebrafish.","publishedAt":"2026-08-27T00:00:00.000Z","fetchedAt":"2026-09-17T13:49:34.664Z","url":"https://www.anthropic.com/news/model-hardware-standard-research-preview","media":[{"type":"image","url":"https://www-cdn.anthropic.com/images/4zrzovbb/website/0aa7623ba0e545ce39e0de725583c0445ce6b76d-1999x1250.png","alt":"The automated experiment workflow","r2Key":"releases/298a48958fec275133ab6076a09931f13da54cc6e86dab06fc06180b074dfb32.png","r2Url":"https://media.releases.sh/releases/298a48958fec275133ab6076a09931f13da54cc6e86dab06fc06180b074dfb32.png"},{"type":"image","url":"https://www-cdn.anthropic.com/images/4zrzovbb/website/943af4b43644ab0d213ba362eabd609c98cd8887-1999x1125.jpg","alt":"Genentech scientists analyze plates for the presence of bubbles","r2Key":"releases/da4cf378e8f4dde71c53b2384f726dd5203dd6b40b661e30eff8518b03fae584.jpg","r2Url":"https://media.releases.sh/releases/da4cf378e8f4dde71c53b2384f726dd5203dd6b40b661e30eff8518b03fae584.jpg"},{"type":"image","url":"https://www-cdn.anthropic.com/images/4zrzovbb/website/34951ed9f87af7ef8b6d639b6e68f4f294e36804-1999x912.png","alt":"Figure 1. Comparing an academic lab, an automated lab, and an MHS-based lab","r2Key":"releases/34a1a691e39bebb83f1f0de1424f589b9ecdd10c3daad851092161e374d88fc0.png","r2Url":"https://media.releases.sh/releases/34a1a691e39bebb83f1f0de1424f589b9ecdd10c3daad851092161e374d88fc0.png"},{"type":"image","url":"https://www-cdn.anthropic.com/images/4zrzovbb/website/b451d157b06b5977035c6c711eea044c89c15f43-1999x1377.png","alt":"Figure 2. Monitoring instruments using MHS","r2Key":"releases/3befd9c8beca2b2d5cdb419ceddea677e4909083f3ff59b12c07ff0cf8e2a58f.png","r2Url":"https://media.releases.sh/releases/3befd9c8beca2b2d5cdb419ceddea677e4909083f3ff59b12c07ff0cf8e2a58f.png"},{"type":"image","url":"https://www-cdn.anthropic.com/images/4zrzovbb/website/cd09961c40c470e184f91a38957262312caa2260-1999x1125.png","alt":"Figure 3. Using an AI agent to monitor and control an experiment in real time via MHS","r2Key":"releases/78596a5cc4dd62002ea60297c43492f17b683ef5d4e18b61bcb622dc89a9cb87.png","r2Url":"https://media.releases.sh/releases/78596a5cc4dd62002ea60297c43492f17b683ef5d4e18b61bcb622dc89a9cb87.png"},{"type":"image","url":"https://www-cdn.anthropic.com/images/4zrzovbb/website/05e6926e8a2eadae5156b8b4601e0e650a55f27b-1464x1074.png","alt":"CMU laboratory instruments","r2Key":"releases/d4d71b7cdb8da01164eb49462628c1ba8c1b2bbd6e63ff0c9a229ca0664dfbe4.png","r2Url":"https://media.releases.sh/releases/d4d71b7cdb8da01164eb49462628c1ba8c1b2bbd6e63ff0c9a229ca0664dfbe4.png"},{"type":"image","url":"https://www-cdn.anthropic.com/images/4zrzovbb/website/4e42afedb6a8c896d966151945c74bbe3fb052dc-1466x1076.png","alt":"CMU laboratory instruments, continued","r2Key":"releases/9bbd7e3d9c34bf71efb8d8e5931bf6518775fedd96bbe87bf7e80038082af0d5.png","r2Url":"https://media.releases.sh/releases/9bbd7e3d9c34bf71efb8d8e5931bf6518775fedd96bbe87bf7e80038082af0d5.png"},{"type":"image","url":"https://www-cdn.anthropic.com/images/4zrzovbb/website/5445017faa35cb3c25fd6b81a610536bfc00c036-1466x1076.png","alt":"A 96-well plate arranged as a serial dilution","r2Key":"releases/ba0c696fd476f8749dae93b11da5e2e7183112ca7d23830d073bb65f5c4b1e6b.png","r2Url":"https://media.releases.sh/releases/ba0c696fd476f8749dae93b11da5e2e7183112ca7d23830d073bb65f5c4b1e6b.png"},{"type":"image","url":"https://www-cdn.anthropic.com/images/4zrzovbb/website/10cf5c1fe1c219338eb7ad8c99fffb70ae2af86e-1639x1019.png","alt":"Run 1. The first serial dilution experiment tested concentrations up to 200 µg/mL"},{"type":"image","url":"https://www-cdn.anthropic.com/images/4zrzovbb/website/3463b88b2a74c3e9111fe1f996e4e58aeca3f45d-1639x1019.png","alt":"Run 2. The system automatically repeated the experiment with a lower maximum concentration of 100 µg/mL"}],"coverageCount":0}],"pagination":{"nextCursor":"2026-08-27T00:00:00.000Z|2026-09-17T13:49:34.664Z|rel_mPCuBFp6D_YFB0mT7gp0e","limit":20},"summaries":{"rolling":{"windowDays":90,"summary":"Anthropic shipped two releases focused on expanding Claude's capabilities for creative and technical work. Opus 4.7 raised performance across coding, agents, vision, and multi-step reasoning tasks. Claude Design emerged as a new Labs product enabling direct visual collaboration—users can now work with Claude to produce designs, prototypes, slides, and one-pagers without leaving the interface.","releaseCount":2,"generatedAt":"2026-04-17T21:23:22.344Z"},"monthly":[]}}