Judge Rita Lin's preliminary injunction, issued Thursday in the Northern District of California, blocked the Pentagon from enforcing its "supply chain risk" designation of Anthropic. The ruling preserved the company's existing government contracts and prevented the Defense Department from barring Anthropic from federal procurement. Headlines across CNN, the BBC, CBS, and NPR framed the decision as a victory. The word "blocks" appeared in nearly every one. [1] [2]

This paper reported Thursday that a judge stopped the Pentagon from blacklisting Anthropic and noted that the precedent mattered more than the case. Two days later, that assessment holds. What has changed is the clarity of what comes next -- and it does not favor Anthropic.

Politico published Friday under the headline "'Premature': Anthropic still in trouble despite court win, lawyers and lobbyists say." The article reported that the AI startup must still convince Trump-appointed judges in the D.C. Circuit Court of Appeals to uphold the pause. The Pentagon's Chief Technology Officer told Breaking Defense the ban "still stands" and that the department is pursuing the appeal on an expedited basis. The supply chain risk designation remains on the books; it is merely enjoined from enforcement. [3] [4]

The legal mechanics explain why victory declarations were premature. A preliminary injunction is not a ruling on the merits. It says only that the plaintiff demonstrated a likelihood of success sufficient to warrant preserving the status quo while litigation continues. Judge Lin's finding -- that the Pentagon's process was "sufficiently irregular" and that the inconsistency between private communications and formal action "raises questions that merit full consideration" -- is strong language for an injunction. But it is not a judgment. The government retains every tool it had before the ruling, and it has added one: the appeal itself, which moves the case from a district court judge in San Francisco to a D.C. Circuit panel where Trump appointees hold a majority. [1] [3]

There are also avenues that bypass the court entirely. The Pentagon retains broad procurement discretion. It can decline to renew Anthropic's existing contracts when they expire. It can exclude the company from future solicitations on grounds unrelated to the supply chain risk label. Federal agencies that receive informal guidance from the White House to avoid Anthropic products will not be deterred by a district court injunction. The formal designation was the most visible weapon. It was not the only one.

OpenAI, which took Anthropic's $950 million classified defense contract within days of the blacklisting, has not been asked to return it. The contract transfer, unlike the designation, was not challenged in court and is not subject to the injunction. The practical effect is that Anthropic won a procedural battle while its largest government deal remains in a competitor's hands.

The X reaction was divided. AI safety advocates and tech-labor accounts celebrated. Policy analysts and defense-tech accounts read the Politico piece and the Breaking Defense reporting more carefully. The emerging consensus on the analytical side of AI X: the court preserved Anthropic's ability to fight, but the fight is just beginning, and the Pentagon sets the venue for the next round.

-- DAVID CHEN, Beijing