The Trump administration is pushing Meta to comply with voluntary safety reviews of its artificial intelligence models, according to sources cited by the New York Times. Federal officials have been urging the lone major holdout among US AI developers to participate in government evaluations. The pressure comes weeks after the administration ordered Anthropic to pull its latest model.
Meta’s reluctance to submit its AI systems for review sets it apart from competitors like OpenAI, Google, and Anthropic, which have already entered into agreements with the government. This divide highlights ongoing tensions between innovation and oversight in the rapidly evolving AI sector. The voluntary program is designed to assess risks before models are deployed widely.
The administration’s push follows a broader effort to impose guardrails on AI development. While the exact terms of any potential agreement remain unclear, the government has signaled that compliance could influence regulatory outcomes. Critics argue voluntary programs lack teeth, while proponents see them as a compromise between speed and safety.
The outcome could set a precedent for how future administrations handle AI oversight. If Meta resists, it may face increased scrutiny from regulators and lawmakers. Other developers will be watching closely, as any agreement could shape industry norms.
Some industry observers question whether voluntary reviews are sufficient to address systemic AI risks. They argue that without binding rules, companies might prioritize competitive advantage over safety commitments.