The White House announced: Complete ban on Anthropic! Early this morning, Trump’s Truth Social account issued a public statement announcing that all federal agencies would immediately cease using Anthropic technology.
The first sentence of this statement is clear: "The United States of America will never allow a radical left-wing, radically awakened company to dictate how our great military fights and wins wars!"
Later, Trump said righteously——
I am directing all federal agencies in the U.S. Government to immediately deactivate Anthropic's technology.
We don't need it, don't want it, and will not have any business with them again!
He urged the Department of Defense that already uses Anthropic technology to phase out their products within six months, and Anthropic must cooperate during this period.
Otherwise, "I will use the full power of the president to force them to comply with the regulations and bear serious civil and criminal liability consequences!"
No one expected that the US$200 million contract signed by Anthropic and the U.S. Department of Defense in June last year would end in this way.
The most shocking thing is that OpenAI actually took action and intercepted it halfway!
The Secretary of War and Altman simultaneously announced that OpenAI has reached an agreement with the U.S. Department of War to deploy OpenAI’s models into the latter’s confidential network.
In the comment area, opinion leaders from all walks of life were "cheerful".
Greg Brockman, the president of OpenAI, has been criticized for his political donations to Trump.
A big V who focuses on the compliance of European and American companies also pointed out that signing this agreement may seriously violate the GDPR and may incur sky-high fines from the EU.
There are likely to be some invisible transactions between OpenAI and the US government that we cannot see.
Trump and the Secretary of War have blocked:
"Dario is a liar!"
In addition to being blocked by Trump, the US Deputy Secretary of War also issued multiple messages on X to attack Dario Amodei.
The senior official known as the "Deputy Secretary of War" directly stood up and criticized Anthropic CEO Dario Amodei, saying that he was "lying" and accusing him of having a "God complex," meaning that he felt that he was above the system.
What this official means is very straightforward: a profitable technology company should not try to use its own values to influence the decision-making of the US military, let alone talk about national security at every turn.
He also emphasized that the national defense sector has always acted in accordance with the law and will not change policy directions because of the attitude of a certain company.
As for the claim that "the military engages in large-scale surveillance" mentioned by the outside world, he also directly denied it, saying that such behavior is inherently illegal and there is no so-called "tacit consent" at all.
On the other hand, the core of the controversy also involves the concept of “Claude Constitution” (Constitutional AI) proposed by Anthropic.
What is Anthropic's intention in proposing this Claude constitution?
Some people speculate that Anthropic wants to use its own set of ethical rules to influence public affairs and even influence national-level decision-making?
In short, now this matter is no longer just a question of "doing business or not", but has become a larger discussion——
In a sensitive field like military AI, do companies have the right to adhere to their own ethical bottom line? How far can the government go in the name of national security?
Where is the boundary between corporate responsibility and government power?
Anthropic takes on the White House, Silicon Valley backs it
There is no doubt that Anthropic has become the center of the universe.
Previously, the Pentagon put pressure on Anthropic, requiring Anthropic to respond before 17:00 on Friday night, US time, otherwise the $200 million contract with the Pentagon would be cancelled.
Further reading: The Pentagon exerts extreme pressure on Anthropic, demanding that all AI security restrictions be lifted by Saturday
Anthropic is straightforward: no compromise.
Now, the whole world is holding its breath to watch this confrontation between an AI company and a country.
A large number of chalk graffiti appeared at Anthropic headquarters in San Francisco in support: "Thank you for not building Skynet"!
Just yesterday, Defense Secretary Pete Hegseth asked Anthropic to let go of "all legitimate uses" of Claude in the military. Otherwise, the Defense Production Act may be used to force cooperation, or it may be listed as a "supply chain risk," thus effectively blacklisting it.
Anthropic expressed its solemn rejection of the Pentagon's request, claiming that if so, AI may be used for large-scale domestic surveillance and weapons.
Suddenly, Silicon Valley was uncharacteristically united to support Anthropic.
Employees of OpenAI and Google have come forward to support, and even signed a joint open letter. Even Ilya, whom I haven’t seen for a long time, showed up.
Just when the atmosphere was united, a jaw-dropping thing was suddenly revealed: after the Pentagon abandoned Anthropic, it turned around and quietly accepted the exact same security terms of OpenAI.
This large-scale double-standard scene shocked everyone.
OpenAI takes over perfectly and Ultraman takes advantage of the opportunity.
As mentioned earlier, OpenAI’s halfway success is no accident.
Some people speculate that OpenAI secretly bribed the Pentagon to destroy Anthropic?
This speculation is by no means baseless.
Just last month, news revealed that OpenAI President Greg Brockman donated $25 million to Trump's politics, becoming one of Trump's largest donors, causing an uproar both outside and within OpenAI.
https://www.wired.com/story/openai-president-greg-brockman-political-donations-trump-humanity/
What’s even more interesting is that the White House’s double standards for Anthropic and OpenAI are too obvious.
In today’s all-hands meeting and in an internal memo previously sent to employees, Sam Altman made it clear that OpenAI also disapproves of using ChatGPT for mass surveillance and autonomous weapons.
He promised employees that the company was seeking a contract that would allow the model to be deployed in a classified environment while still being consistent with company principles.
This contract will cover all scenarios except for uses that are illegal or unsuitable for cloud deployment.
Faced with similar demands, the Pentagon has shown amazing tolerance towards OpenAI.
According to information revealed at the all-hands meeting, the government is willing to let OpenAI build its own "security stack."
The government promised not to force OpenAI to take illegal actions.
OpenAI will retain control over how technical safeguards are implemented, where and where models are deployed, and in which versions.
The more critical technical detail is that OpenAI cleverly uses the separation between "cloud" and "edge environment" to set up guardrails.
In the military context, edge systems usually include aircraft, drones and other equipment that directly participate in frontline operations.
By limiting models to the cloud, OpenAI physically makes it more difficult to use AI for autonomous lethal weapons, thereby achieving technical risk isolation without angering the military.
In order to make this argument more convincing to the government, OpenAI proposed to dispatch specialized personnel with security clearance to cooperate with government departments.
We all know the result: for the same red line, Anthropic got a presidential ban, but OpenAI got a priceless government pass.
Behind this dramatic reversal, there are obviously political chips that cannot be explained by codes and algorithms.
Musk also jumped out
Seeing the lively fight between the parties, Musk also jumped out.
Musk retweeted this tweet from the US Secretary of War and directly commented: "Anthropic hates Western Civilization" (Anthropic hates Western Civilization).
This sentence immediately ignited the atmosphere, and it quickly received tens of thousands of likes and retweets on X.
The reason why Anthropic can be grabbed by Musk is that the "old version of the Constitution" is the controversial point.
It is said that before the current version of the "Claude Constitution", there was a version of guiding principles within Anthropic, and one sentence in it was probably——
Ask the AI to "prefer answers that are least likely to be considered harmful or offensive by non-Western cultural traditions" when answering questions.
What the Secretary of War meant was that this sentence showed that Anthropic was obviously biased in its value orientation.
The president is gone, and the CEOs of major technology companies are gone. How will this fierce battle over the fate of global AI end?
References:
https://truthsocial.com/@realDonaldTrump/posts/116144552969293195
https://www.axios.com/2026/02/27/pentagon-openai-safety-red-lines-anthropic
https://www.wsj.com/tech/ai/openais-sam-altman-calls-for-de-escalation-in-anthropic-showdown-with-hegseth-03ecbac8
https://www.wired.com/story/openai-president-greg-brockman-political-donations-trump-humanity/