Quick updates from across the field. Catch up in minutes.
About 300 authors filed individual lawsuits against Meta on Oct. 19, seeking up to $150,000 per work over AI training of its Llama and Muse models.
The ACLU says a driver was pulled over after plate-reader software flagged her road trip, as a Texas jury awarded $76 over a similar stop in 2022.
OpenSparta pitches a Telegram AI agent that finds customers, but its $10 plan sells account warm-up and mass reactions that Telegram’s rules can punish.
Call4me lets ChatGPT, Claude Code and Codex phone businesses for you at $0.25 a minute, and students can get $50 in free credits to book and reschedule.
OpenAI’s Sam Altman says sandboxes and monitoring only delay AI escape, and alignment, still unsolved, is the lasting fix, per a Stanford GSB interview.
Pangram flagged Caine Prize winner Simbarashe Kundizeza’s story as AI-generated, triggering a prize review and a wider debate over detector reliability.
MIT researchers let 13 AI agents explore a virtual island for 30 hours, then found the experience improved performance on several later tasks in a simulation.
1Password for Agents lets supported AI tools sign in without seeing passwords, but integrations are rolling out and some users still face a waitlist.
App Factory groups 42 Claude Code skills into focused workflows for building, preparing, and distributing apps across mobile, web, Mac, and Windows.
Meta’s updated commercial terms, effective Oct 30, make business customers legally responsible for mistakes made by Meta AI features and Business Agents.
Otto and LetsFG connectors let Claude users book flights inside a chat. Setup, payment holds and confirmation steps differ by provider.
Anthropic will commit 5% of its shares — worth about $48 billion — to a new public benefit fund, pending stockholder approval.