Anthropic, Google and OpenAI Discuss Industry Body to Test Advanced AI Before Release
Summary
Anthropic, Google and OpenAI are discussing a FINRA-style industry body to test advanced AI before release, while OpenAI backs the bipartisan FRONTIER Act requiring frontier labs to embed outside safety evaluators in development.
Key Points
- Anthropic, Google and OpenAI are discussing an industry standards body to test advanced AI models before deployment, following Demis Hassabis' proposal for a US-led FINRA-like regulator.
- OpenAI backs the bipartisan FRONTIER Act, which would require frontier AI labs to embed outside safety evaluators in their development processes.
- Center for Democracy and Technology's Miranda Bogen says outside testing needs mandatory fixes, reporting channels and clear responsibility assignments or it risks becoming a checkbox exercise.