The Trump administration is asking Meta (designated as an extremist organization and banned in Russia) to voluntarily submit its artificial intelligence models for government evaluation, the New York Times reported on Tuesday, citing four informed sources.
According to the publication, Facebook’s parent company is the only major U.S. AI developer that has yet to agree to voluntarily hand over its models to the federal government for review. In April, Meta unveiled its Llama artificial intelligence model—the most advanced of its creations to date.
The government review will enable officials to assess the capabilities and vulnerabilities of the AI models.
Earlier this month, Washington mandated that Anthropic restrict foreign nationals’ access to its most advanced AI models, citing security concerns. Anthropic complied with the requirement by fully barring public access to the models.
OpenAI and Anthropic have previously worked with the U.S. government on testing unreleased AI models. Google DeepMind, Microsoft, and xAI have also recently agreed to undergo national security reviews.
About the Author
