- Lab-grown clumps of human brain cells can be wired up and trained to play simple video games, and researchers say larger versions would raise animal welfare questions; the piece argues they undercut the assumption that sentience is an on-or-off trait and the hope that AI would preserve humans for their unique brains (Mitch)
- A Massachusetts bill that has cleared the state Senate would require frontier AI firms to open their practices to outside evaluators, protect whistleblowers, and report serious safety incidents to the state (Joe)
- OpenAI and Anthropic have hired lobbyists on opposite sides: OpenAI wants a lighter Illinois-style standard and says reviews would delay cybersecurity models, while Anthropic backs the bill on the grounds that companies should not grade their own work (Joe)
- The bill's weak points include no public funding for evaluators, which leaves developers paying them, and a start date of January 2028 (Joe)
- Reporting on AI and inequality notes that spending by wealthier Americans with AI stock gains is holding up the economy, while workers in India earn $2.50 an hour recording their daily chores on head-mounted cameras so robots can learn those tasks (Alana)
<- Back to digests
Daily digest
Uncomfortable topics
By Mitchell Howe, Joe Rogero, and Alana Horowitz Friedman
Human brain organoids, AI company proxy fight, and wealth inequality
Read the full dispatch on AI StopWatch
From AI StopWatch, published with their permission.
The summaries and the full Spanish translation are produced automatically. The original is always linked.