Monday, June 29
TL;DR Anthropic’s June Economic Index landed today: half of 9,700 surveyed Claude users say AI can already handle 50%+ of their job. OpenAI’s GPT-5.6 Sol/Terra/Luna remain gated to ~20 US-approved orgs under the new frontier AI review process.
Models & research
-
Anthropic published the June Economic Index “Cadences” report today: among ~9,700 surveyed Claude users, 50% say AI can handle half or more of their job tasks today, 4% say it could replace their entire role now. Work queries taper on weekends, high-income tasks cluster at night. model-releases Anthropic · GIGAZINE
-
OpenAI previewed GPT-5.6 as three tiers on June 26 (Sol flagship at $5/$30 per 1M tokens, Terra balanced at $2.50/$15, Luna cheap at $1/$6) but limited access to ~20 trusted partners at the US government’s request under Trump’s June 2 frontier AI EO; broad rollout expected in coming weeks. model-releases OpenAI · CNBC
People
- simonw ran a public red-team on June 26: 2,000 people made 6,000+ attempts to extract a secret from his Claude-powered AI assistant; zero succeeded. His read: less worried about prompt injection than before, but would not deploy this pattern for anything irreversible in production. simonwillison.net · HN thread
Tools & dev
- claude-code Dynamic Workflows ships in research preview: Claude Code can spawn hundreds of parallel sub-agents for codebase-scale migrations, security audits, and architecture reviews in one session. Token costs are “substantially more” than a standard session. Available on Max, Team, and eligible Enterprise plans. Anthropic blog · InfoQ
OSS
msitarzewski/agency-agentstops GitHub trending today (1,221 stars) — Shell-based AI agency framework with specialized expert agents for various task domains. agentic-ai GitHubbrowser-use/video-use(976 stars today) — video editing through coding agents, automates production workflows end-to-end. agentic-ai GitHubHKUDS/Vibe-Trading(840 stars today) — autonomous personal trading agent framework in Python. GitHub