> And they have “heavily” deployed Musk’s Grok AI chatbot – an aspiring ChatGPT rival – as part of their work slashing the federal government, said that person. Reuters could not establish exactly how Grok was being used.

Grok is FIPS complaint, right? Safe for use with confidential information? PII/PHI?

> Grok is FIPS complaint, right? Safe for use with confidential information? PII/PHI?

“Laws without enforced consequences are merely suggestions.” -- Ron Brackin

From what we've seen so far in this admin, none of that matters much anymore.

Compliance? What's that? Sounds too lawful. Must be a relic of the past.

Because ofcourse they do... But really, what is the way to be biggest AI company ever, eclipsing palantir and openai? Yeah, make sure the US and other govs believing they cannot function without anymore. I see where Musk is going, but like his self driving cars, he is always a just not quite there yet. And this is even more dangerous than self driving cars that suck.

I am a bit like him; when I make something, I see the end goal clearly and tell people like it's already there, but it doesn't exist yet and takes years to do. Difference is, I sit in my garden in my underpants, he stands ranting in the white house.

Not one individual quoted in the article. Not 1.

Seems like legacy media is not pretending anymore and has transitioned into make-believe land after CNBCs airing of "unconfirmed information" about tariff pause only 1 day ago.

I'm sure you mean that the article doesn't name any federal worker or person with knowledge of the events which they've quoted or paraphrased. In which case, of course. This administration has shown it will vindictively punish any dissent or disloyalty. Nobody wants to give their name for that. I will not disbelieve accounts just because the sources won't attach their names to them. If the administration wants better coverage, they need to accept that their employees may not always agree with or like them.

Some quotes from the article (was it really that hard to search for quotation marks yourself?):

>"We have been told they are looking for anti-Trump or anti-Musk language,” a third source familiar with the EPA said.

>"Be careful what you say, what you type and what you do,” a manager said, according to one of the sources.

>Last year, before Trump was elected, Musk suggested AI could be used to replace government workers, according to a person with direct knowledge of his comments. “The concept was that through taking the government data that they could build the most dynamic AI system ever,” the person said, adding that AI could then “do the work.”

… Hold on, are you expecting them to name their sources? On this? Really?

Like, protecting sources is about as old as journalism. Without it, you’re really just a conduit for press releases.

Unfortunately bad faith actors also use "anonymous sources," to push agendas that aren't exactly an accurate representation of what's going on. The FBI and CIA are notorious for this.

Example. "Anonymous sources in the CIA said..." That's a leak but not a real leak. You can tell because if it was a real leak, the government would track them down like Snowden.

is the issue surveillance, or AI?

> is the issue surveillance, or AI?

Both.

I'm sure one of the biggest eventual use cases of AI will be surveillance at scale. If that's the case, how enthusiastic are you be about AI as a technology?

The issue is a lack of accountability. The technology is irrelevant.

Yup - people should contact their representatives.

Yes. Surveillance because his weird organization doesn’t fit anywhere in actual org charts and its unclear who plays oversight for it or handles its data security, and because AI is notoriously prone to hallucinations and shouldn’t be used deterministically.

Why not both?

There is no expectation of privacy but there is a need for rules and oversight. Some communications are sensitive - anything involving lawyers, unions, health providers, etc. has legal considerations - and there is both a concern about whether government data is being shared improperly with third-party companies. For example, if they’re using Grok to analyze text, there’s potentially a huge conflict of interest if any of the communications being analyzed involves regulation of Musk’s companies or the various lawsuits he’s part of.

Similarly, since they’re creating a conflict between perceived political disloyalty and professional ethics you have questions like whether an EPA official whose messages are flagged will have a human review the alleged offenses or be given a chance to defend their actions, or simply be “randomly” included in some RIF. We haven’t had to think about that since the McCarthy / lavender scare era, which significantly predated modern surveillance technology.

Since DOGE is widely reported to be using Signal and private email servers, there is reason to question whether those ethical standards will be followed.

I have no problem with surveillance of federal workers on their job sites, per se. I say this as a federal worker. (Edited for clarity.)

As others have mentioned, there are two legitimate concerns here:

(1) Civil servants are heavily trained on the laws restricting use and access to certain information. There's reasonable doubt that DOGE and/or Trump will uphold those laws with these systems additional surveillance systems in place.

(2) This administration has proven itself incompetent and criminal regarding HR activities. Civil servants can reasonably wonder if this will further those dysfunctions.

> I have no problem with surveillance per se.

OK. For the rest, we have elections and in the interim contact your congresspeople. Believe it or not, some percent of the country supports these actions, so the only solution is to remove them through the democratic process and put pressure on elected officials to hold them accountable.

You don't have to assume I'm American. But yeah, the officials that interpret law at the highest level should absolutely be democratically elected. Why would that even be some sort of gotcha?

it's not a gotcha - it's just that what you're saying is already how it is for most of the judicial branch, so it's not really a big deal.

DOGE, or alternatively Elon Musk.

And even if one can argue whether those count as foreign entities as in other countries, they're certainly foreign to the US government, since this is quite novel BS.

Wasn't Musk appointed by Trump though?

It's pretty sparse with details and could just be speculation, but it sounds like "sentiment analysis at scale". Traditional monitoring would be keywords or manual oversight, the former being ineffective and the latter being labour intensive. Having AI monitoring every communication for wrong-think, where wrong-think is simply arguing against a direction or initiative, is...novel.

There is something rather interesting in Trumpism where everyone has to fall in line in ways never, ever seen before. The dictator dictates and everyone else starts parroting. There is incredible danger if no one in government is allowed to express disagreement.

Reuters and AP used to be pretty much newswires, but lately they seem to act more like a newspaper.

> Reuters could not establish exactly how Grok was being used.

> So, this article is a "we don't actually know" and a "they could be doing something but, but again we don't know".

> 0/10 Reuters

What do you want? The design specification and implementation? By what seems to be your standard, every newspaper would be empty. Which would be very convenient to some people! For instance, Elon Musk and DOGE.

Reuters knows what their sources are telling them, and they're reporting that. They're doing a fine job.

> Trump administration officials have told some U.S. government employees that Elon Musk's DOGE team of technologists is using artificial intelligence to surveil at least one federal agency’s communications for hostility to President Donald Trump and his agenda, said two people with knowledge of the matter.

I think its entirely reasonable for an employer to monitor employee communications on company media channels. I've dealt with a few toxic employees in the past where their incessant complaining and passive sabotage of work is incredibly harmful to morale and work of others.

I don't really see how using AI is controversial other than it's being done more efficiently, flagging just blatant violations. Seems preferable and more respectful of privacy that having an employee read all communications manually and applying judgement.

Has an LLM ever gotten something wrong in your experience? Now imagine you get fired because the magic autocomplete machine randomly decided that something you said had a bad vibe

Trump isn't their employer and the whole narrative that they "report" to him is an executive order-based fiction. The departments are supposed to be independent, some wildly more than others, and the heads of the department control them and set the agenda.