US presses Meta to submit AI models for security reviews
What's new: Meta is the only major US AI developer without a voluntary review deal as Washington expands national-security checks.
The Trump administration is pressing Meta to let the federal government review its AI models before wider release, as Washington tightens oversight of frontier AI over national-security risks. The New York Times reported that officials asked the Facebook parent in confidential emails to join a voluntary program that evaluates models' capabilities and vulnerabilities. Meta said it shares the goal of secure AI and hopes to sign an agreement soon. OpenAI and Anthropic were already testing unreleased models with the government, while Google DeepMind, Microsoft and xAI agreed in May to provide early access for security evaluations. Trump on June 2 created a framework for developers to offer covered frontier models to the government for as long as 30 days before release to trusted partners.