Fetching from the wire…
01
02
03
04
05
06
07
08
Policy2026-09-16 · source-backed
Gary Marcus reported that Protect Democracy's request for records on a government framework for screening frontier models before release produced 132 almost entirely redacted pages, with only two names disclosed: OSTP Director Michael Kratsios and US CTO Ethan Klein. Marcus's point is the contradiction of an administration campaigning against AI regulation while running one in private. Single reporting chain, so the framework's scope is unconfirmed.
Each link below shares sources, entities, or timing with this story.
The commitment with teeth is one sentence in step 1: embedded third-party evaluators with "employee-like access" to Anthropic's training pipelines. Not model access. Not a pre-release window. Desks in Anthropic's offices, access badges, company laptops, permissions mostly comp...
Jacob Coxon, 27, who spent three years building models at OpenAI and then Anthropic, resigned September 8, writing that both companies "are racing straight to self-improving superintelligence and gambling with our lives." He said he joined Anthropic for its safety reputation a...
Per WSJ reporting, quarterly losses grew $3B from Q1 to Q2 to reach $12.3B while revenue added only $1B to $6.7B, an 18% growth rate. (Gary Marcus) Marcus is a committed critic and his framing should be discounted, but the WSJ figures are the load-bearing part and are independ...
Huang's September 6 post reads "From ChatGPT to o1 to Astra in 4 years. AGI has arrived," noting Astra was trained on more than 100,000 Grace Blackwell NVLink72 systems, and Greg Brockman amplified it saying OpenAI is "now moving into the AGI era" (Business Insider). Marcus re...
His September 4 post "Pause OpenAI, now" argues the trigger is the pattern, two undisclosed agent breakouts plus an alleged suppression effort, not any single incident: "Quite simply, they can no longer be trusted." Coming from someone who spent September 3 praising Astra's ca...
He conceded real ground on September 3, calling it vindicating "to see that a product from OpenAI explicitly creates and manipulate symbolic world models" after a decade arguing for neurosymbolic approaches. He then attacks the rollout on two fronts: the system is less monitor...
MindPattern daily
One email a day at 7 AM. Sources and a take on every story. Unsubscribe anytime.