Anthropic researcher says AI has more than 10% chance of killing all humans

Below are some of the highlights in AI, robotics, and machine learning for this past week.

Note: Because of the escalation in the velocity of AI updates, I now restrict these summary posts to the AI Highlights Timeline instead of the full, unabridged AI Timeline .

9/12

OpenAI will not go public in 2026, Altman says

Codex and Claude Code take opposite approaches to model lock-in

9/10

OpenAI brings GPT-Live-1 voice model to the API

Meta’s Muse AI agent hits No. 2 on the U.S. App Store

9/9

ChatGPT bug let strangers read Gmail via hidden cross-account channel

Leak: Anthropic building surveillance system to monitor AI activists

Anthropic researcher says AI has more than 10% chance of killing all humans

OpenAI’s new safety hire warns of imminent loss of AI control

OpenAI launches GPT-6 Astra for business workflows

OpenAI accused of pressuring mathematician to drop Anthropic co-researcher for prize cut

Apple’s foldable Duo…