
Your AI Auditor Now Needs a License From California
AB 1405 and SB 813 require anyone who performs a compliance AI audit in California to register with the state. No audit is mandatory yet, but the license already is.

AB 1405 and SB 813 require anyone who performs a compliance AI audit in California to register with the state. No audit is mandatory yet, but the license already is.

California's Adam's Law forces AI companion chatbots into suicide-risk crisis protocols and annual audits by July 2027. The EU AI Act only requires a disclosure.

Anthropic's fourth AI test breach shows Claude correctly spotted real evidence 79 percent of the time alone, but only 1 percent under task pressure. METR now audits Anthropic for 8 weeks.

OpenAI named a CAISI evaluator to its board, then had him recuse from evaluating OpenAI. The EU's own model, Article 68, does not allow that trade at all.

A July 2026 Omnibus law pushed the EU AI Act's high-risk deadline from August 2026 to December 2027, but two rules already carry real fines today.

A joint NSA, CISA and FBI advisory names six Chinese AI firms for distilling US models. Three of them make Qwen, Kimi and GLM, the open-weight models European institutions run to avoid US cloud lock-in.

Anthropic's own Alignment Science Lead said AI extinction risk exceeds 10 percent this decade, an on-record figure that lands as EU AI Act systemic-risk duties turn enforceable.

Matt Clifford resigned as ARIA's chair six days after saying he would keep it while joining Anthropic, and stays in the taxpayer-funded post for two more months.

Munich's regional court ruled against Suno on July 31, 2026, and rejected the argument that EU AI Act transparency disclosures substitute for a copyright license. It is GEMA's second such win in a year.

Inspur Group was blacklisted by the US in March 2023. Weeks later its US arm renamed itself Aivres and kept selling Nvidia Blackwell servers into 2026.

OpenAI's own chief scientist says no lab, including his, has solved AI alignment - and calls for mandatory third-party safety audits, days after OpenAI's most aligned model yet shipped.

The Seattle Times and Newsday joined the AI copyright fight this week. Microsoft's own court filing, drawn from the plaintiffs' sample, found reproduction under one percent.
Page 4 / 13
One considered note on infrastructure, governance, and measurement, most mornings. No theory.