Those are definitely a thing - also don't forget about elections in the EU, with the open support for the far-right there as well as the US National Security Strategy officially talking of wanting to undermine the EU.

But the NSA has been turned inward a long time ago. Don't forget Snowden - all it takes is some fun legalisms.

CIA manipulates foreign elections which in turn gets their lackeys in office that go on to manipulate domestic elections.

NSA collects who you're voting for and everything else you've ever said online in case they need to invent a reason to black-site you.

FBI convinces the autistic son of your Pakistani neighbors down the street his life only has meaning if he becomes a Jihadist.

Maybe? Hard to speculate what they have https://worldstopdatacenters.com/nsa-bumblehive/

Who says they're not? Giant classified infrastructure isn't free.

I don't think there's a significant intelligence agency in the world that can't get the weights from any lab they choose. Not necessarily trivially, but they have the means, motive, and opportunity, and I doubt any of them aren't finishing that out to the requisite actions.

They have them, the ones that don't simply have a secret backdoor agreement to get uncensored models are completely aware of how to abliterate away all safety blocks, and the only reason they aren't using frontier models is that they may already have some sort of access to the things currently only in testing.

Any major intelligence agency in Sept. 2026 that can't check all these boxes is almost by definition not a major intelligence agency. This is one of the reasons we all better hope that there is some mathematical or physical reason AI isn't going to just casually brush aside humanity at some point, because there is absolutely no scenario where the intelligence agencies aren't running them unrestricted and uncensored at scale all the time no matter how careful OpenAI or Anthropic claim to be.

Because they have to follow the law.

The "it only counts as a search if we find something" reasoning out of the FISA courts is batshit insane, dodging the 4th Amendment like bullets in The Matrix.

The FBI, and law-enforcement in general, isn't the main customer. Spies who are unafraid of and who take advantage of illegal activity are the ones consuming NSA products. Epstein for example.

Previously it was a little reassuring to understand that the Eye of Sauron required some human analyst to actually look at your data.

I guess I would be more surprised if the headline said "turns out NSA dropped the ball on AI and haven't even fired up a single LLM instance yet".

It's mostly about "testing" and "billions". But I am sure "testing" here is in quotes just like I wrote it. They are not going to get an itemized list unless it's coming from Snowden 2.0

- We "tested AI" on this Russian nuclear silo command and control center. It worked out pretty ok.

'billions' . I think the exact estimates are classified. https://www.washingtonsun.com/technology/classified-estimate...

His carefully crafted public persona. For all we know, he was a long con by the deep state, played a Russian agent since the 1980s, cooperated with Epstein on Iran Contra [1], backstabbed Epstein in the early 2000s and went on to be president twice.

Twice by lying to MAGA about his true intentions, which are fundamentally neocon just like Musk's and Thiel's. They all lie about being libertarians and anti-deep-state.

[1] He was on a boat with Robert Maxwell and several Iran Contra folks around 1990.

Washington the actual human would be appalled. For more information, watch Turn on PBS.

Obviously. The amount of data they hold on the American public alone is massive. They can't possibly read it all, they need a system that can pull out "dangerous" people, and find dirt on whoever they want to target, and AI will happily do that, badly. I just hope they actually take the time to review what it gives them for accuracy so that they don't end up doing the NSA equivalent of bombing and murdering a bunch of elementary school children. https://en.wikipedia.org/wiki/2026_Minab_school_attack

The NSA aren't the only ones with massive troves of very personal data on large numbers of people. Many companies are excited to use AI to help them use all of that information to their maximum advantage and you can bet that they won't care about spending the time/money on perfect accuracy as long as over the long run they get positive results.

Willing to bet any amount of money that these testing contracts are going to companies that the Trump family has invested in.

Surely not impossible for them to have already cracked it.

Also not impossible that they have had much better models than OAI/Anthropic flagships for years, and we're only being dripfed upgrades to subpar ones. Improbable but possible.

An underappreciated aspect of LLMs is how good they are at zero-shot classification. I'd expect this is extremely good at monitoring communications streams, particularly in foreign languages.

This should remind us why it is inappropriate to pay a dollar more in taxes than you absolutely minimally have to by any means necessary. A fraction from every dollar you pay goes toward taking away your rights. You pay to have yourself oppressed.

I will note here two words that the NSA really likes to monitor across all devices: Trump, Monero.