Anthropic vs. DoW #6: The Court Rules
TL;DR
Judge Lynn didn’t just pause the government — she torched its whole theory — In a 43-page ruling, she granted Anthropic a preliminary injunction, saying the company is likely to win on First Amendment retaliation, due process, APA, and statutory grounds, including the line that nothing in the law supports the “Orwellian notion” that a U.S. company can be branded a saboteur for disagreeing with the government.
Anthropic’s strongest factual win was simple: the parties were still negotiating when the government labeled it a risk — A March 4 email from Under Secretary Emil Michael to Dario Amodei said “we are very close here” and showed the last live dispute was wording around domestic surveillance, badly undercutting the claim that Anthropic had become some urgent supply-chain threat.
The government’s technical case collapsed because Anthropic can’t secretly ‘pull the plug’ — Sworn declarations from Sarah Heck and Thiago Ramassamy said there is no backdoor, no kill switch, no unilateral update path, and compared model deployment to a cake recipe: once served, neither diner nor chef can change it at the table.
Emil Michael’s own declarations, as presented here, mostly reduce the case to ‘Anthropic had red lines and asked annoying questions’ — The narrator argues Michael effectively admitted the risk designation was triggered by Anthropic insisting contract terms have force of law, refusing all-lawful-use language, and publicly criticizing the government’s position.
Judge Lynn treated the blacklist as real-world economic damage, not just a technical paperwork fight — The ruling cites major law firms warning clients to audit Anthropic exposure and says within days the company saw deals worth hundreds of millions delayed, customers pull out, and contracts terminated.
A bizarre late government angle — Anthropic employs foreign nationals — was framed as both weak and dangerous — The video notes every major frontier AI lab relies heavily on foreign talent, with Samuel Hammond citing Meta’s 5,123 approved H-1B petitions in FY2025 and estimating foreign nationals make up roughly 50–65% of frontier AI research teams.
The Breakdown
The ruling lands, and the host is basically delighted
The video opens with the big news: Anthropic got its preliminary injunction, stayed for seven days, and Judge Lynn is portrayed as both furious and unusually clear-eyed. The host says this is one of the most devastating judicial opinions he has ever read and sounds genuinely giddy that the judge understood the core issues without needing deep AI expertise.
Anthropic’s evidence says the government changed course mid-negotiation
The recap then rewinds to Anthropic’s response brief and the supporting declarations from Sarah Heck and Thiago Ramassamy. The standout exhibit is Emil Michael’s March 4 email saying “we are very close here,” which the host treats as devastating because it suggests the parties were still hashing out narrow wording — especially around surveillance — right before Anthropic was declared a supply chain risk.
The technical rebuttal: no kill switch, no sabotage, no magical model drift
Anthropic’s declarations are presented as a point-by-point demolition of the Pentagon’s implied technical fears. The memorable analogy is the cake: once the model is deployed into a government environment, Anthropic can’t reach in and tweak the recipe, and any new version would require review and approval. The host lingers on this because it turns a vague fear of “they might pull the plug” into something physically impossible.
The foreign-worker argument makes the case look thinner, not stronger
Next comes a side path the host clearly finds absurd: the Department of War arguing Anthropic’s foreign workforce is a security risk. He and quoted sources like Samuel Hammond argue this logic would torch the whole U.S. AI sector, since companies like Meta, Google, Microsoft, and OpenAI all rely heavily on foreign talent, and Anthropic is actually known for stronger-than-average insider-threat controls.
Judge Lynn’s pre-hearing questions signaled she saw through the game
Before the hearing, Lynn issued pointed questions that the host calls brutal: Did Hegseth’s tweet have legal effect? Why didn’t the designation discuss less intrusive alternatives? How exactly could Anthropic sabotage anything after delivery? The hearing itself, as recapped here, went badly for the government, especially when it tried to say the tweet meant nothing while also refusing to clearly undo the damage it caused.
Emil Michael files declarations and, in this telling, mostly incriminates himself
Michael’s declarations are the next major beat, and the host goes through them with open contempt. His summary is that Michael’s case boils down to: Anthropic had contract red lines, insisted those terms mattered, questioned one operation, and therefore became “untrustworthy.” The host’s running point is that this isn’t evidence of sabotage risk — it’s evidence of retaliation for speech and for refusing bad contract terms.
Judge Lynn drops the hammer in language that practically glows
Then comes the payoff: the ruling itself. Lynn says the measures look designed to punish Anthropic, cites public statements calling the company “out of control” and “arrogant,” and says punishing a company for bringing public scrutiny to government contracting is classic illegal First Amendment retaliation. The host is especially animated by the ruling’s tone, repeatedly emphasizing that she found the government likely unlawful on almost every available theory.
What happens next: appeal, more litigation, and the risk of escalation
The injunction blocks enforcement of the presidential directive, the Hegseth directive, and the supply chain risk designation, while still allowing the government to simply stop buying Anthropic if it wants. The case now heads to the Ninth Circuit, where the host expects the government to appeal anyway, partly because backing down would look worse politically. His closing mood is triumphant but wary: Anthropic just scored a huge win, but the real unresolved question is whether the government chooses to de-escalate or retaliate further.