Grok, Elon Musk’s scandal-plagued artificial intelligence platform, has burrowed deep into the federal government—and federal officials have some concerns!
Too bad, says President Donald Trump. He just banned all government agencies from working with one of Musk’s biggest competitors, Anthropic, even though its main AI product, Claude, is already extensively deployed in military systems.
It’s little coincidence that the reason Grok is in is the same reason Claude is out.
A few days ago, Musk’s AI company xAI signed an agreement that Grok could be used in classified military systems. Previously, only Anthropic’s Claude could be used on things like weapons development and battlefield ops. But Anthropic wouldn’t agree to the Department of Defense’s ultimatum that it remove guardrails and limitations on Claude.
Related | Hegseth sets alarming conditions for AI company's contract
You no doubt have guessed that Musk has no issues with having no guardrails. Musk happily signed the demand that Grok could be used for “all lawful use,” a terrifyingly vague term, particularly in an administration that does not at all respect the rule of law. Do you really want the lawful use of your product defined by Defense Secretary Pete Hegseth?
The problem for the people who have to use Grok and, apparently, no longer use Claude is that Grok is a shitty AI platform.
On Friday, only a few hours before Trump’s temper tantrum, the Wall Street Journal reported that officials at several agencies are concerned about Grok’s safety and reliability, and that it is—brace yourself—“sycophantic and too susceptible to manipulation or corruption by faulty or biased data.
Even the top official at the General Services Administration, Ed Forst, raised concerns about Grok. Now, Forst is no career civil servant. Instead, he’s a former private equity executive, so if he’s freaking out, it’s bad.
But Josh Gruenbaum, a senior official at the GSA who was recruited by Musk’s so-called Department of Government Efficiency, assuaged White House Chief of Staff Susie Wiles that Grok’s government platform is separate from its scandal-plagued public one. And apparently, that was good enough for Wiles.
All the high-level love of Grok is not shared by the people who have to use it. The Journal characterized the demand as “anemic,” with a source telling the outlet that Grok had been used only in a few instances where people needed to “mimic a bad actor for defensive testing.”
Well, earlier this year, the chatbot was creating nonconsensual sexual images on demand, so that definitely seems like it can play the role of “bad actor.”
All of this Grok adoration played out as the government essentially went to war with Anthropic over its desire to maintain some guardrails on the use of Claude, forbidding the military from using it for mass domestic surveillance and autonomous weapons.
Related | Oh great, Elon Musk's pervy AI is about to pollute the Pentagon
In response, Undersecretary of Defense Emil Michael rage-posted on X late Thursday that the head of Anthropic “wants nothing more than to try to personally control the US Military and is ok putting our nation’s safety at risk,” and that the DOD will “not bend to whims of any one for-profit tech company.”
Buddy, settle down. The government’s existing contract with Anthropic included those guardrails. But then Whiskey Pete rolled in last month and said all contracts now have to be amended to the “all lawful use” standard, which contradicts the limitations in the contract the parties signed.
Since the government has no interest in adhering to its contract, and since Anthropic seemingly withstood the administration’s usual bullying, it changed tactics.
Earlier this week, Hegseth threatened Anthropic that the government would invoke the Defense Production Act, which would give the military the authority to use Claude however it wants, regardless of what Anthropic wants. However, the Pentagon also threatened to designate Anthropic as a “supply chain risk,” meaning anyone who wants to do business with the military could not have any ties to or use of Anthropic and Claude.
As Lawfare pointed out, it isn’t clear how Claude can simultaneously be so vital to the interests of the United States that it had to be used in all defense efforts, no matter what, but also so dangerous that no one doing business with the federal government could use it.
But President Trump has solved everything in his very special way: by yelling on Truth Social, because this is how government works now:
THE UNITED STATES OF AMERICA WILL NEVER ALLOW A RADICAL LEFT, WOKE COMPANY TO DICTATE HOW OUR GREAT MILITARY FIGHTS AND WINS WARS! That decision belongs to YOUR COMMANDER-IN-CHIEF, and the tremendous leaders I appoint to run our Military. The Leftwing nut jobs at Anthropic have made a DISASTROUS MISTAKE trying to STRONG-ARM the Department of War, and force them to obey their Terms of Service instead of our Constitution.
Can someone ask Trump where in the Constitution it says that private companies with government contracts have to let the government do things explicitly forbidden by the contract?
We must have missed that one.
Oh well. Enjoy having Grok loose in all your most sensitive systems, DOD! That will surely go well for all of us.
Lisa Needham February 27, 2026 at 11:30PM From Daily Kos
0 Comments