OpenAI, Google and Meta Pledge Outside AI Audits Under Voluntary White House Deal | CoinDesk Daily
Thursday, 1 October 2026 · 1 min read · Listen to the episode ↗
OpenAI, Google, Meta, Anthropic, Nvidia and xAI agreed to let outside auditors review safety controls for their most capable AI models under a voluntary White House pact with no deadline or penalties. The episode also examines the crypto industry’s more than $8 million in first-half lobbying for a market-structure bill that failed, and CFTC proposals that could bolster federal oversight of prediction markets.
OpenAI, Google, Meta, Anthropic, Nvidia and xAI agreed to a voluntary White House pact allowing outside auditors to review their AI safety controls. The companies are asked to monitor their most capable models for the potential to enable cyberattacks or biological and chemical threats.
President Donald Trump called the pact “morally binding,” but it sets no deadline and carries no penalties. The commitments therefore depend on voluntary participation rather than an enforcement mechanism specified in the deal.
The crypto industry spent more than $8 million lobbying Congress in the first half of the year for a digital assets market structure bill that ultimately failed. Coinbase spent about $2.2 million, putting it among the top 10 securities and investment lobbying spenders and above Goldman Sachs.
An unnamed former SEC official pointed to significant internal infighting and a lack of unity on key decisions as obstacles in the crypto bill’s progress.
The CFTC submitted two proposed rules to the White House that would change its approach to disputes with states over prediction markets. One would define event contracts as swaps under federal jurisdiction; the other would declare them not to be gambling.
If both proposals take effect, they could weaken state lawsuits alleging that platforms such as Polymarket and Kalshi operate illegal gambling businesses.
This summary was generated from the episode transcript and can contain mistakes.