Anthropic, Google and OpenAI have discussed setting up an AI industry standards body. People familiar with the talks described them to CNN. The trigger was a July essay by Google DeepMind founder Demis Hassabis. He proposed a US-led body modelled on the Financial Industry Regulatory Authority. It would test advanced models before release, be overseen by government and funded by industry. The talks are ongoing with or without the Trump administration. Treasury Secretary Scott Bessent was reported in July to be weighing a similar agency reporting to the SEC. Meta CEO Mark Zuckerberg advised Trump against a national AI regulator. House Speaker Mike Johnson said the companies have no consensus on guardrails. The White House already runs an unpublicised voluntary pre-release review.
What changed
Few industry standards exist for how AI models are tested before release.
- voluntary review up to 30 days pre-release
Sources