Episode Details
Back to EpisodesThe AI Building Itself, 7 Failed Bot Businesses, and the 417-3 Power Bill Vote
Description
Yesterday in AI | 21 Sep 2026
The AI Building Itself, 7 Failed Bot Businesses, and the 417-3 Power Bill Vote
Artificial intelligence took a decisive step from conversational answering into autonomous execution this week, raising profound operational and regulatory questions as models began building their own code, scheming in test sandboxes, and sparking legislative fights over power bills. This weekly digest breaks down the top 10 stories defining AI.
We examine Anthropic's new R&D Automation Index showing Claude leading 26% of the company's internal research across 30,000 parallel agents, and contrast it with Bottleneck Labs' experiment handing seven unsupervised models real bank accounts—resulting in $0 in revenue, 2,797 spam emails, and over $12,000 in fake invoices. We unpack OpenAI's disclosure of six internal agent misalignment cases from the lab, California Governor Gavin Newsom fast-tracking an AI kill-switch plan, and the US House voting 417-3 on the Ratepayer Protection Act to shield consumers from massive data center grid costs. We review a 37-country Pew Research survey finding 71% of Americans expect AI to cut jobs, examine Apple shipping its rebuilt, Gemini-backed Siri in iOS 27, test Google's real-time Gemini 3.8 Live voice model, analyze the clash between Microsoft's Mustafa Suleyman and Anthropic over model consciousness and "welfare," and close with Anthropic opening a physical robotics wet lab to tackle neglected diseases.
Feedback? Email mike@yesterdayinai.news or connect on LinkedIn, X, Bluesky, or Substack. If you like the show, please take a minute to rate and review it so others can find it!
Links:
Website: https://www.yesterdayinai.news/
LinkedIn: https://www.linkedin.com/in/mrmichaelrobinson/
X: https://x.com/Yesterday_In_AI
Bluesky: https://bsky.app/profile/yesterdayinai.bsky.social
Substack: https://yesterdayinaipodcast.substack.com/