Trump plan to combat AI risks hinges on Big Tech pals policing themselves

President Trump announced a voluntary accord signed by roughly two dozen AI firms to carry out external safety audits and adopt controls outlined by the White House. The public summary, which Trump posted on Truth Social, emphasizes independent testing of firms’ internal controls, monitoring, and detection systems with attention to cybersecurity, biosecurity, chemical threats, and unintended model behaviors. The deal requires companies to meet regularly to share best practices and develop common safety standards and benchmarks, but it does not create new legal obligations for signatories. Several companies named in the accord—including Anthropic, Google, and OpenAI—had already publicly committed to external audits, and some firms are participating in confidential government safety testing programs. Details about how the promised external reviews will be carried out were limited in the announcement. Sources cited by Bloomberg and The Information said it is unclear who will serve as third-party auditors and that the White House’s selection process appears politicized to some observers. The Center for AI Standards and Innovation, which runs voluntary pre-deployment government tests, has reportedly met with some firms that could conduct reviews. Critics described the pact as largely symbolic. Some industry insiders told Axios they worry that self-policing won’t solve the safety problems that have prompted pauses in model training and releases. Public figures, including Senator Bernie Sanders, criticized the White House’s embrace of industry leaders; meanwhile, Trump labeled the agreement “morally binding” and hinted the recommendations could be codified later. The announcement came as poll data cited by Bloomberg showed strong public support for stricter AI guardrails.