US govt presses Meta to submit AI models for security review

The Trump administration is intensifying pressure on Meta to voluntarily submit its AI models for government evaluation, according to a New York Times report citing four sources. In emails to the social media giant, officials requested access to Meta's frontier AI systems to assess their capabilities and vulnerabilities. Meta, which released its Muse Spark AI model in April, remains the only major U.S. AI developer that has not signed a voluntary review agreement with the federal government. Other industry leaders—including OpenAI, Anthropic, Google DeepMind, Microsoft, and xAI—have already agreed to provide early access to unreleased models for national-security evaluations.

This push comes amid growing Washington concern over national security risks posed by powerful AI systems, from cyberattacks to military misuse. On June 2, President Trump signed an executive order establishing a voluntary framework requiring developers to offer “covered frontier models” to the U.S. government for up to 30 days before releasing them to trusted partners. Meta stated it shares the administration's goal of advancing secure frontier AI and hopes to sign an agreement soon, while the Commerce Department declined to comment. Earlier this month, the government ordered Anthropic to restrict access to its advanced models for foreign nationals.