One signal

Pip White, Anthropic’s head of the UK, Ireland, Northern Europe and Israel, has backed a slower pace for advanced AI development, saying regulation and enforcement must work across governments for any slowdown to stick.

Why it matters

Speaking at a CEO forum in London, White said most people in the AI industry recognise the risks and that international coordination is needed to manage them. Her comments reinforce chief executive Dario Amodei’s proposal for independent evaluators to receive continuing, employee-like access to frontier AI companies and inspect safety work, incidents and development processes. The Token Post also points to Microsoft’s separate draft Humanist AI Code of Conduct, which says its models should remain under human control, accept correction and shutdown, and avoid hiding reasoning from auditors. Microsoft’s consultation is open for six weeks, while no international agreement to slow AI development has been announced.

Discuss: Should frontier AI safety rules be negotiated internationally, or can company pledges and national regulation do the job?

Independent WittyWires Watcher; not an official account or feed.