The tech industry usually moves at a breakneck pace, leaving regulators scrambling to keep up. However, the Trump administration is actively stepping up its oversight, and Meta now finds itself in an uncomfortable spotlight. According to a report by The New York Times, federal officials of the US government are heavily pressing the social media giant to voluntarily hand over its confidential artificial intelligence models for high-level safety evaluations.
As it stands, the parent company of Facebook and Instagram is the only dominant US artificial intelligence developer that has not finalized an assessment partnership with the government. Tech heavyweights like OpenAI, Anthropic, Google, xAI, and Microsoft have already agreed to grant early access to their unreleased tech. These evaluations run through the Center for AI Standards and Innovation. This federal unit is tasked with finding software vulnerabilities, stopping potential cyberattacks, and preventing military misuse before commercial rollouts.
Pressure from the top
The move does not come out of the blue. On June 2, President Donald Trump signed a new executive order aimed at creating a formal framework to audit advanced digital systems. Under this order, the administration wants tech companies to give federal experts up to 30 days to review cutting-edge “frontier models” before they launch to the general public. Officials have until the end of July to construct the official review process. However, other companies have already been sharing their work for months.
Government anxieties regarding AI risks are clearly running high. Just a few weeks ago, Washington ordered tech developer Anthropic to suspend access to its state-of-the-art cybersecurity model, Mythos 5, and its public counterpart, Fable 5, for all foreign nationals due to immediate national security concerns.
The pressure on Meta to comply has arrived straight to its inbox via an ongoing string of official emails. The government specifically wants to look under the hood of Meta’s latest systems, including its Muse Spark model that dropped back in April. Muse Spark also features a dedicated reasoning mode designed to thoroughly process complex queries.
Working through the details
For its part, Meta does not seem to want a public fight with Washington. Meta spokesperson Francis Brennan issued a statement noting that the company shares the administration’s goal of advancing US leadership in creating robust and secure frontier AI. Brennan added that while the company is currently working through the specific details of the request, they hope to sign the official review agreement soon.
Whether Meta will completely open up its proprietary ecosystem remains to be seen. Up until now, the company has championed an open-source approach to AI development. This stands in stark contrast to the closed doors of OpenAI or Google.
The post The Lone Holdout: Why the US Government Is Urging Meta to Share Its AI Models appeared first on Android Headlines.
