{"collection":{"slug":"frontier-ai-labs","name":"Frontier AI Labs"},"digest":{"id":"cwd_yfOhfYfE00Ff4NDlpX59L","weekStart":"2026-06-15","title":"Health intelligence, agent safety, and enterprise controls","intro":"GPT-5.5 Instant brought frontier health evaluations to all free ChatGPT users with a 71% reduction in factuality issues, while Claude Code hardened agent safety with destructive command blocking and Anthropic launched enterprise-managed MCP auth. OpenAI also shipped credit usage analytics for Enterprise admins and granular spend controls.","body":"### Health intelligence reaches free users\n\nThe week's most impactful release was [GPT-5.5 Instant's improved health intelligence](/release/rel_KxYssNJpkq3VSDeSkDDta-openai-launches-gpt-5-5-instant-with-improved-health-intelligence), now available to all free ChatGPT users. OpenAI reports that production factuality issues in health responses have fallen by 71% over the past two months, and physician-rated responses now surpass both older models *and* human-written responses across accuracy, communication, and completeness criteria. With over 230 million weekly health and wellness queries on ChatGPT, this means a substantial cohort of users now receives materially better answers for interpreting lab results, understanding insurance, or preparing for appointments — no subscription required. The release signals a broader shift: health is becoming a primary evaluation frontier for general-purpose assistants, not just specialized medical AI.\n\n### Agent safety and hardening\n\nClaude Code continued its aggressive safety push with [v2.1.183's destructive git command blocking](/release/rel_LaiiNBn31J-auck7UeRv_-claude-code-v2-1-183-blocks-destructive-git-commands-and-fixes-terminal). Commands like `git reset --hard`, `git checkout -- .`, `git clean -fd`, and `git stash drop` are now blocked unless the user explicitly asked to discard local work, and `terraform destroy`/`pulumi destroy`/`cdk destroy` are similarly gated. This is a direct response to one of the most common failure modes in AI coding agents: a well-intentioned but catastrophic command that obliterates hours of work. The same release fixed fullscreen TUI corruption on Windows Terminal under heavy subagent load — a quality-of-life fix that affects anyone running complex multi-agent workflows on Windows.\n\nThe safety tightening extended to [Claude Code v2.1.178](/release/rel_nsWTtdZ40Ab7bhf7_XMJ9-claude-code-v2-1-178-fixes-oauth-credential-races-and-stale-fd-crashes), which introduced wildcard parameter matching for tool permission rules — you can now write `Agent(model:opus)` to block Opus subagents, for example — and fixed an OAuth credential race that could crash the process via stale file descriptors. Meanwhile, [Codex CLI 0.142.0](/release/rel_iB9NDXJegllWqhyfrOyAg-codex-cli-0-142-0-adds-token-budgets-web-search-and-restores-tui-after-ctrl-z) added configurable token budgets that track usage across agent threads and abort turns when exhausted, plus an indexed web-search mode that restricts direct page access to server-approved URLs, giving teams control over what the agent can read.\n\n### Enterprise admin tooling matures\n\nBoth labs shipped meaningful enterprise controls this week. [OpenAI launched credit usage analytics and granular spend controls for ChatGPT Enterprise](/release/rel_vPxjRzo7HpRkJX2nM8NzL-openai-launches-credit-usage-analytics-and-granular-spend-controls-for-chatgpt), letting admins view usage trends across users, products, and models in the Global Admin Console, and set default, group-specific, or individual spend limits. On the SDK side, both the [Python](/release/rel_lM6e5Mmr8XL9KLkZGY1a_-openai-python-sdk-v2-42-0-adds-admin-spend-alerts) and [Go](/release/rel_4RUazLImEAWDGjFAtYfFu-openai-go-sdk-v3-40-0-adds-admin-spend-alerts-api) SDKs added admin spend alerts API support. [Anthropic launched enterprise-managed auth for MCP connectors](/release/rel_at3la8999w7wVJEBeB-B7-anthropic-launches-enterprise-managed-auth-for-mcp-connectors), allowing admins to authorize and authenticate connectors across the whole organization through their identity provider — no individual setup required.\n\nTwo more channels for Slack integration arrived: [ChatGPT Business](/release/rel_0BC5pMPjOYnDR-YhYmTXf-chatgpt-business-adds-slack-connector-actions-with-oauth-scope-approvals) and [Enterprise & Edu](/release/rel_LAAqAZ0FYJL0QU9WlapPa-chatgpt-enterprise-edu-adds-slack-connector-actions-with-oauth-scope-controls) workspaces can now use Slack connector actions — joining channels, creating reminders, uploading files — with OAuth scope controls managed by workspace admins. Record & Replay also launched as a [Codex feature for macOS Business users](/release/rel_g687WbysikLsyIWhbe82y-chatgpt-business-adds-record-replay-in-codex-for-macos), letting teams demonstrate a workflow once and turn it into a reusable skill (initially outside the EU, UK, and Switzerland).\n\n### Platform updates and the ecosystem\n\nAnthropic's [SDK support for code execution tool](/release/rel_lo3p_kB6eSXfEZO2zMHJc-code-execution-tool-sdk-support-and-web-search-fetch-improvements-june-18-2026) `code_execution_20260120` rolled out across Python, TypeScript, Go, Java, Ruby, PHP, and C# — no beta header required — available on Claude Fable 5 through Sonnet 4.5+. [Claude Code Artifacts](/release/rel_xQCotY3-oSs6ZFWpBulya-claude-code-artifacts-beta-turns-raw-output-into-shareable-visual-pages) entered beta for Team and Enterprise plans, turning raw output into shareable visual pages. [Codex 0.141.0](/release/rel_3En_04jxOrvrbTNlL_XGY-codex-0-141-0-encrypts-remote-executors-adds-plugin-mcp-servers-fixes-sqlite-wal) encrypted remote executors via Noise relay channels and pinned bundled SQLite to fix a WAL-reset corruption bug. And the [MCP tunnels management API migration](/release/rel_PrciwodHzMZG5I7PKF31h-mcp-tunnels-management-api-migration) moved from the Admin API to the Claude API with a new beta header and workspace-level IAM scope — a breaking change with a migration window for existing users.","releaseIds":["rel_KxYssNJpkq3VSDeSkDDta","rel_LaiiNBn31J-auck7UeRv_","rel_nsWTtdZ40Ab7bhf7_XMJ9","rel_iB9NDXJegllWqhyfrOyAg","rel_vPxjRzo7HpRkJX2nM8NzL","rel_lM6e5Mmr8XL9KLkZGY1a_","rel_4RUazLImEAWDGjFAtYfFu","rel_at3la8999w7wVJEBeB-B7","rel_0BC5pMPjOYnDR-YhYmTXf","rel_LAAqAZ0FYJL0QU9WlapPa","rel_g687WbysikLsyIWhbe82y","rel_lo3p_kB6eSXfEZO2zMHJc","rel_xQCotY3-oSs6ZFWpBulya","rel_3En_04jxOrvrbTNlL_XGY","rel_PrciwodHzMZG5I7PKF31h"],"releaseCount":33,"generatedAt":"2026-07-11T16:46:11.869Z","releases":[{"id":"rel_KxYssNJpkq3VSDeSkDDta","title":"OpenAI launches GPT-5.5 Instant with improved health intelligence","path":"/release/rel_KxYssNJpkq3VSDeSkDDta-gpt-5-5-instant-health-performance-matches-frontier-models","org":{"slug":"openai","name":"OpenAI"},"importance":null},{"id":"rel_LaiiNBn31J-auck7UeRv_","title":"Claude Code v2.1.183 blocks destructive git commands and fixes terminal corruption","path":"/release/rel_LaiiNBn31J-auck7UeRv_-destructive-git-commands-blocked-fullscreen-tui-corruption-fixed","org":{"slug":"anthropic","name":"Anthropic"},"importance":null},{"id":"rel_nsWTtdZ40Ab7bhf7_XMJ9","title":"Claude Code v2.1.178 fixes OAuth credential races and stale FD crashes","path":"/release/rel_nsWTtdZ40Ab7bhf7_XMJ9-stale-fd-crash-fixed-oauth-account-mismatch-handled","org":{"slug":"anthropic","name":"Anthropic"},"importance":null},{"id":"rel_iB9NDXJegllWqhyfrOyAg","title":"Codex CLI 0.142.0 adds token budgets, web search, and restores TUI after Ctrl+Z","path":"/release/rel_iB9NDXJegllWqhyfrOyAg-token-budgets-and-indexed-web-search-tui-survives-suspension","org":{"slug":"openai","name":"OpenAI"},"importance":null},{"id":"rel_vPxjRzo7HpRkJX2nM8NzL","title":"OpenAI launches credit usage analytics and granular spend controls for ChatGPT Enterprise","path":"/release/rel_vPxjRzo7HpRkJX2nM8NzL-credit-usage-analytics-and-granular-spend-controls-for-enterprise","org":{"slug":"openai","name":"OpenAI"},"importance":null},{"id":"rel_lM6e5Mmr8XL9KLkZGY1a_","title":"OpenAI Python SDK v2.42.0 adds admin spend alerts","path":"/release/rel_lM6e5Mmr8XL9KLkZGY1a_-admin-spend-alerts-added","org":{"slug":"openai","name":"OpenAI"},"importance":null},{"id":"rel_4RUazLImEAWDGjFAtYfFu","title":"OpenAI Go SDK v3.40.0 adds admin spend alerts API","path":"/release/rel_4RUazLImEAWDGjFAtYfFu-admin-spend-alerts-api-ships","org":{"slug":"openai","name":"OpenAI"},"importance":null},{"id":"rel_at3la8999w7wVJEBeB-B7","title":"Anthropic launches enterprise-managed auth for MCP connectors","path":"/release/rel_at3la8999w7wVJEBeB-B7-mcp-connectors-authorized-through-identity-providers","org":{"slug":"anthropic","name":"Anthropic"},"importance":null},{"id":"rel_0BC5pMPjOYnDR-YhYmTXf","title":"ChatGPT Business adds Slack connector actions with OAuth scope approvals","path":"/release/rel_0BC5pMPjOYnDR-YhYmTXf-slack-actions-join-remind-upload-in-chatgpt-oauth-scope-management","org":{"slug":"openai","name":"OpenAI"},"importance":null},{"id":"rel_LAAqAZ0FYJL0QU9WlapPa","title":"ChatGPT Enterprise & Edu adds Slack connector actions with OAuth scope controls","path":"/release/rel_LAAqAZ0FYJL0QU9WlapPa-slack-connector-actions-with-oauth-scope-controls","org":{"slug":"openai","name":"OpenAI"},"importance":null},{"id":"rel_g687WbysikLsyIWhbe82y","title":"ChatGPT Business adds Record & Replay in Codex for macOS","path":"/release/rel_g687WbysikLsyIWhbe82y-record-replay-turns-workflows-into-reusable-skills","org":{"slug":"openai","name":"OpenAI"},"importance":null},{"id":"rel_lo3p_kB6eSXfEZO2zMHJc","title":"Code execution tool SDK support and web search/fetch improvements (June 18, 2026)","path":"/release/rel_lo3p_kB6eSXfEZO2zMHJc-code-execution-tool-sdk-support-and-web-search-fetch-improvements-june-18-2026","org":{"slug":"anthropic","name":"Anthropic"},"importance":null},{"id":"rel_xQCotY3-oSs6ZFWpBulya","title":"Claude Code Artifacts beta turns raw output into shareable visual pages","path":"/release/rel_xQCotY3-oSs6ZFWpBulya-artifacts-beta-raw-output-becomes-shareable-visual-pages","org":{"slug":"anthropic","name":"Anthropic"},"importance":null},{"id":"rel_3En_04jxOrvrbTNlL_XGY","title":"Codex 0.141.0 encrypts remote executors, adds plugin MCP servers, fixes SQLite WAL corruption","path":"/release/rel_3En_04jxOrvrbTNlL_XGY-remote-executor-channels-now-e2e-encrypted-wal-reset-corruption-fixed","org":{"slug":"openai","name":"OpenAI"},"importance":null},{"id":"rel_PrciwodHzMZG5I7PKF31h","title":"MCP tunnels management API migration","path":"/release/rel_PrciwodHzMZG5I7PKF31h-mcp-tunnels-management-api-migration","org":{"slug":"anthropic","name":"Anthropic"},"importance":null}]}}