Early leaks about President Trump’s executive order on AI made it sound like a crackdown that would have forced companies to let the U.S. government vet AI models for safety before they could be released to the public. But it didn’t take long for those behind the scenes to signal what was really coming: voluntary vetting.
The executive order he signed on June 2 asks companies to submit their new models for government oversight. Tech companies should, pretty please, submit to a 30-day model evaluation period, and the government gave itself until July 31 to nail down a review process.
But so far, Meta doesn’t wanna, according to an anonymously sourced story in the New York Times.
The Times notes that while Anthropic, OpenAI, Google, xAI, and Microsoft have signed on to be vetted by the Commerce Department’s Center for A.I. Standards and Innovation (CAISI), Meta hasn’t, and not because White House isn’t trying.
It’s notable that Anthropic is acting like a model citizen toward the White House considering it just had its flagship model forced offline by the White House’s actions. That company is also still designated a supply chain risk by the Pentagon, and Trump’s Defense Secretary, Pete Hegseth, has shown no sign that he plans to back down about that.
Nonetheless, it’s Meta who seems to be the last major holdout. Multiple emails have apparently been sent urging voluntary compliance. Meta spokesman Francis Brennan wrote in a statement to the Times that he and his company “share the administration’s goal of advancing U.S. leadership on robust and secure frontier A.I.,” and that while they are “working through the details,” they “hope to sign the agreement soon.”
Meanwhile, a Commerce Department spokesman named Ben Kass told the Times, “This story is not unusual,” and that it’s “the very work CAISI is supposed to be doing.”
Read the full article here
