Today's briefing →

This Week's Synthesis

Distrust that began with Nvidia's governance spread to Anthropic's listing and a math-proof claim, and the AI industry did not solve the problem of checks lagging behind big claims this week

Nvidia was at the center of the daily insight from Monday to Thursday. Petition signatures for Jensen Huang's return passed 180,000, yet the board drew a line against it even as it posted record results. Meanwhile Nvidia halted shipments to China, delayed its quarterly report and attached conditions to neocloud guarantees. What the market reacted to was not profit but trust in the board and the investigations.

That distrust moved to Anthropic, which listed on the 19th. On its second trading day, the 20th, Anthropic's stock fell 8.5% to within five cents of its $83 IPO price, and the same day Anthropic mathematician Levent Alpöge announced that he and Claude had proved the Riemann hypothesis. An all-human competing draft from Stanford appeared hours later, and neither proof has been accepted by mathematicians.

On Sunday, an analysis found that AI-discovered vulnerabilities passed 100,000 this year, leaving companies little time to validate patches, and the encrypted messenger Signal restricted screenshots to block AI agents. What the week had in common is that in stocks, proofs and security patches alike, the claim and the speed arrived before the checks.

This Week's Daily Briefings

Past Weekly Recaps