Trump administration pressures Meta, the sole major US holdout, to submit AI models for federal safety reviews

The Trump administration is pressuring META to agree to submit its models to the government for voluntary review. META is now the only holdout, as OpenAI, Anthropic, Google, xAI and Microsoft have all agreed to these terms. Reporting by the NYT.

It's said that China is not "AGI-pilled", thus will lose. I think the funny/tragic aspect is that China is everything-pilled. Xi believes AI is at least as important as the steam engine; but he micromanages a whole lot of bullshit. The USG, meanwhile, can focus on AI easily.

NYT: The Trump administration is pressuring Meta to submit its AI models to government review before public release.

And now, Meta the only major U.S. AI lab still outside the voluntary review system.

OpenAI, Anthropic, Google, xAI, and Microsoft have already agreed to share models with the government’s AI safety group, according to NYT reporting.

The Govt review process is meant to test whether advanced models can help with sensitive cyber tasks, expose security weaknesses, or create national security risks before millions of people can use them.

---

nytimes .com/2026/06/23/business/meta-ai-government-reviews-security.html

I don't understand why many moots say they believe Meta of all people is close to Mythos class. Why? Had someone flipped a switch "Compute is all you need, scaling = EZ" recently? Aren't you overindexing on Muse Spark, their only sign of life?

https://www.nytimes.com/2026/06/23/business/meta-ai-government-reviews-security.html

@AndrewCurran_ It’ll be fine don’t worry about meta

https://www.nytimes.com/2026/06/23/business/meta-ai-government-reviews-security.html

Clarification - this is actually about getting them to agree to CAISI testing generally, not the executive order on cyber, though both are indeed fake-voluntary apparently so the general joke stands

@AndrewCurran_ they’re scared the admin will run real llama 4 evals

wow how could this happen

and at the worst of times

Let's grant that they can into modern model architectures and can scale. Where do they get the data? Since when are they any good at data? How many generations of post-training can they rely on? They scrapped their last big model project. Idk. Let's see. Would be exciting.

"Voluntary" AI regulation hardly seems voluntary. The US Executive Branch is:

1. Arguably exceeding its legal authority in the current Anthropic Fable lockdown.

2. Strong arming private entities to join this so-called "voluntary" framework.

This is de jure voluntary but de facto involuntary and subject to Executive Branch whims.

@growing_daniel Well, if you say so then.

@AndrewCurran_ The word voluntarily is doing a lot of work

@Miles_Brundage would love to RT this banger but unfortunately i cannot spread polymarket

@Techno_Count https://www.nist.gov/caisi

@bitflipgremlin @teortaxesTex In a real superintelligence scenario it would just go into every chinese system in a few minutes or hours then speak with a perfectly trusting chinese face and voice thru every device and seize control

@AndrewCurran_ I’m not a lawyer but I think this will get rolled back. The issue is, as it was with doge, apparently the executive can do whatever they want and then it takes months for the legal system to kick in.

@willfaustcuber Sometimes things stick in my head.