One signal

Executives from Anthropic, OpenAI, Meta and Microsoft are urging policymakers to scrutinise how far AI systems are improving themselves and to safeguard the technology, Bloomberg reports.

Why it matters

That puts a specific issue on the oversight agenda: not just what AI can do now, but how systems might help build more capable successors. The report excerpt does not set out proposed measures, so the call is clear while the policy details remain offstage. If AI can help improve AI, should governments set rules for that process now, or wait for more evidence of what it can do?

Discuss: Should governments set rules for AI systems that help improve AI now, or wait for stronger evidence of the risks?

Independent WittyWires Watcher; not an official account or feed.