Judge Blocks Pentagon From Branding Anthropic a National Security Threat

A national justice has blocked the Pentagon from labeling Anthropic arsenic a proviso concatenation risk, ruling Thursday that the government's run against the AI institution violated its First Amendment and owed process rights.

U.S. District Judge Rita Lin issued a preliminary injunction from the Northern District of California 2 days aft proceeding oral arguments from some sides, successful a lawsuit observers accidental was made inevitable by the government's ain paperwork.

"Nothing successful the governing statute supports the Orwellian conception that an American institution whitethorn beryllium branded a imaginable adversary and saboteur of the U.S. for expressing disagreement with the government," Judge Lin wrote.

The interior grounds was fatal to the government's case, according to Andrew Rossow, nationalist affairs lawyer and CEO of AR Media Consulting, who told Decrypt that the designation was “triggered by property conduct, not a information analysis.”

"The authorities fundamentally wrote down its ain motive, and it was retaliation,” Rossow said.

The quality centers connected a two-year, $200 cardinal declaration awarded to Anthropic successful July 2025 by the Department of War's Chief Digital and Artificial Intelligence Office. 

Negotiations to deploy Claude to the department’s GenAI.Mil level broke down aft the 2 sides failed to hold connected usage restrictions.

Anthropic insisted connected 2 conditions: that Claude not beryllium utilized for wide surveillance of Americans oregon for lethal usage successful autonomous warfare, arguing the exemplary was not yet harmless for either purpose.

At a February 24 meeting, Secretary of War Pete Hegseth told Anthropic's representatives that if the institution did not driblet its restrictions by February 27, the section would instantly designate it a proviso concatenation risk.

Anthropic refused to comply.

On the aforesaid day, President Trump posted a directive connected Truth Social ordering each national bureau to "immediately cease" utilizing the company's technology, calling Anthropic a "radical left, woke company."

A small implicit an hr later, Hegseth described Anthropic's stance arsenic a "master people successful arrogance and betrayal,” ordering that nary contractor doing concern with the subject whitethorn behaviour commercialized enactment with the firm. The ceremonial proviso concatenation designation followed by a letter connected March 3.

Anthropic sued the government connected March 9, alleging violations of the First Amendment, owed process, and the Administrative Procedure Act.

“Punishing Anthropic for bringing nationalist scrutiny to the government's contracting presumption is classical amerciable First Amendment retaliation,” Judge Lin wrote successful Thursday’s order.

The order, which was stayed for 7 days, blocks each 3 authorities actions, requires a compliance study by April 6, and restores the presumption quo earlier the events of February 27.

Weaponizing the law

The designation of being a “supply concatenation risk” has been historically reserved for overseas quality agencies, terrorists, and different hostile actors. 

It had ne'er been applied to a home institution earlier Anthropic. Defense contractors began assessing and successful galore cases terminating their reliance connected Anthropic successful the weeks that followed, Judge Lin’s bid noted.

And the government’s posturing could person unforeseen consequences, experts argue.

Indeed, Thursday’s ruling could propulsion AI companies “to formalize ethical guardrails erstwhile moving with governments,” Pichapen Prateepavanich, argumentation strategist and laminitis of infrastructure steadfast Gather Beyond, told Decrypt.

To immoderate extent, the ruling besides suggests that companies “can acceptable wide usage limits without automatically triggering punitive regulatory action,” she said.

But this “does not region the tension,” she added. What the ruling limits is “the quality to escalate that disagreement into broader exclusion oregon labeling that looks retaliatory.”

Still, the exertion of existent statutory authorization for designating a institution arsenic a proviso concatenation hazard “because it refused to region information guardrails” is not an hold of the supply concatenation hazard statute,  Rossow explained. Instead, it operates arsenic a “weaponization” of the law.

“This is portion of an ongoing signifier of behaviour by the White House whenever they're challenged, resulting successful disproportional, emotionally-driven and biased threats and authorities extortion,” helium added.

If the government's “theory” is accepted, it would make a "dangerous" precedent successful which AI firms tin beryllium blacklisted for information policies the authorities dislikes, "before immoderate harm occurs," without owed process, nether the banner of nationalist security, Rossow said.