A weekly roundup of the AI news we covered this week.
- China says it now runs 70+ embodied-AI training grounds— China reports 70+ embodied-AI training grounds in use, plus a ministry push to move embodied AI from labs to scaled deployment.Cyberspace Administration of China
- China's November internet summit gets an open-source theme— The 2026 Wuzhen Summit (Nov 2-5) will run 27 sub-forums and put open-source model makers and frontier AI safety research at the centre.Cyberspace Administration of China
- Anthropic pairs a critical-infrastructure security programme with a free open-source scanner— Anthropic is pairing a critical-infrastructure defence programme with OSS Scanner, a free security scanner for open-source projects.Anthropic
- Google DeepMind extends watermarking to AI-designed proteins— DeepMind's SynthID Bio watermarks AI-generated protein sequences and structures, verified in wet-lab tests.Google DeepMind
- NIST signs on to a White House science push framed around super intelligence— NIST backs a White House science initiative, pledging open datasets and a role in the Genesis Mission Consortium.NIST