Meta Resists AI Lock

Meta is reportedly facing intense pressure as the sole major tech giant holding out on sharing its AI models with the US government for safety reviews, according to the New York Times. While competitors like OpenAI, Anthropic, Google, xAI, and Microsoft have already agreed to grant early access to their unreleased systems, Meta has yet to finalize its own agreement.

This notable delay comes as the government aggressively steps up its national security oversight, recently ordering Anthropic to block all foreign nationals from accessing its advanced Mythos-5 and Fable-5 models. Under a June executive order, officials are racing to establish a formal review process by the end of July, aiming to secure a 30-day window to evaluate new AI technologies before they launch.

Meta, which released its Muse Spark model with a reasoning-capable thinking mode back in April, has been receiving these government requests via email. A Meta spokesperson disclosed that they share the goal of securing Frontier AI and hope to sign the review agreement soon, leaving the company in a race to comply as Washington tightens its grip on the industry.