US Urges Meta to Share AI Models for Review

The US government is urging Meta to voluntarily share its artificial intelligence models for evaluation due to escalating safety and security concerns. According to reports, Meta is the sole major AI developer that has not yet submitted its models for government review. This initiative aims to assess AI capabilities and identify potential vulnerabilities.

Companies such as OpenAI, Anthropic, Google, xAI, and Microsoft have already agreed to provide early access to their new models to a government-established center for AI standards and innovation. President Trump's executive order, signed on June 2, mandates the development of a federal review process for AI releases by the end of July, with the goal of granting authorities up to 30 days for technology evaluation before public release.

Meta's recent AI model, Muse Spark, launched in April, features 'Instant' and 'Thinking' modes, with the latter enhancing reasoning capabilities. While Meta acknowledges the goal of advancing secure AI and hopes to sign an agreement soon, the government's requests have been ongoing. Separately, the government has also ordered Anthropic to suspend access to certain models for foreign nationals, citing national security concerns.