On June 12, the Trump administration gave Anthropic ninety minutes to disable its flagship Claude Fable 5 and Mythos 5 models. The order, justified by a “verbal” and “narrow” jailbreak claim, halted a flagship American technology three days after its launch — and exposed once again a presidency that mistakes impulse for strength, retaliation for governance, and grievance for national security.
On Friday afternoon, June 12, 2026, at 5:21 p.m. Eastern, the leadership of one of America’s most important AI companies received a letter from Commerce Secretary Howard Lutnick. By the time most Americans sat down for dinner, Anthropic — a company valued in the tens of billions, with a confidential IPO filing already with the SEC — was forced to “abruptly disable” the two most powerful AI models it had ever shipped. Claude Fable 5 had been public for ninety-six hours. Claude Mythos 5, used by cybersecurity defenders to find and patch vulnerabilities before attackers could exploit them, was killed alongside it. According to reporting confirmed by Axios, the White House gave Anthropic ninety minutes between phone call and shutdown.
This is not a story about one bad policy decision. It is a story about a presidency that no longer governs through deliberation, statute, or institutional process, but through ultimatum and retribution. It is also — and this is the part the country has been too tired or too afraid to say plainly — a story about a man whose decision-making has become so erratic that a bipartisan group of physicians, two senators, fifty House Democrats, and a growing chorus of constitutional scholars believe the Twenty-Fifth Amendment of the United States Constitution must finally be discussed in earnest.
The Anthropic shutdown is the through-line. Examine how it happened, who it hurts, and what it reveals, and the constitutional question stops being abstract.
1. The Friday Order
Anthropic released Claude Fable 5 — its first broadly available model in what the company calls the “Mythos class” — on June 9. The company called it the most capable model it had ever made generally available, with breakthroughs in software engineering, scientific research, and long-running autonomous tasks. A more constrained version, Mythos 5, was offered to a vetted group of “cyberdefenders and infrastructure providers” tasked with finding and fixing software flaws before bad actors could exploit them. Foreign Policy described the launch as Anthropic’s biggest in years.
Three days later, that launch was over. The Trump administration’s export-control directive ordered Anthropic to block all access to Fable 5 and Mythos 5 by any foreign national — anywhere in the world, including Anthropic’s own foreign-national employees inside the United States. Because there is no realistic way for a cloud-based AI product to wall off “foreign nationals” in real time, the order was effectively a global kill switch. Anthropic complied, and disabled the models for everyone.
The administration’s stated rationale was a single jailbreak. According to Anthropic’s official statement, the government provided “only verbal evidence” of “a potential narrow, non-universal jailbreak, which essentially consists of asking the model to read a specific codebase and fix any software flaws” — a capability, the company noted, already available in other publicly deployed AI systems including OpenAI’s GPT-5.5. No written technical brief. No statutory finding. No notice-and-comment process. A phone call, a letter, ninety minutes.
“We disagree that the finding of a narrow potential jailbreak should be cause for recalling a commercial model deployed to hundreds of millions of people. If this standard was applied across the industry, we believe it would essentially halt all new model deployments for all frontier model providers.”
— Anthropic, official statement, June 13, 2026
White House AI adviser David Sacks offered a different account on X, claiming that the administration had reluctantly acted only after Anthropic CEO Dario Amodei refused to fix the flaw or pull the model himself. Semafor subsequently reported that the deeper concern was a suspicion — never disclosed to Anthropic — that a China-linked group had accessed Mythos. Two narratives. One outcome: an American company’s flagship product, deployed to hundreds of millions of users, gone in an afternoon.
2. The Cost — to Workers, Markets, and the National Interest
Conservatives once treated the sudden deployment of state power against a private American company as the defining feature of authoritarian economies. That instinct has now been inverted entirely. The Friday order did not just inconvenience a tech company. It rippled through the American economy in ways the administration appears not to have considered — or to have considered and not cared.
The damage is not confined to one company. American AI sits at the center of the global technology economy. When the U.S. government can summarily revoke the international availability of an American product without a hearing, without a written finding, without anything resembling regulatory due process, every foreign customer and partner of every American technology company learns the same lesson at once: this is a country where access is granted by presidential mood. That lesson is already being absorbed in boardrooms in London, Tokyo, Berlin, and Bangalore. The cost will not be measured this quarter; it will be measured in the contracts that quietly migrate to Mistral, DeepSeek, or whatever the next non-American frontier lab becomes.
For the average American — the engineer whose tools went dark, the small-business owner whose customer-service automation broke, the retiree whose 401(k) sits heavy with tech ETFs — this is not an abstract dispute. It is a direct cost imposed by a chaotic White House, on people who never had a voice in the decision.
3. A Pattern, Not an Incident
To treat the Anthropic shutdown as an isolated event is to misread the situation. It is the latest entry in a documented sequence of impulsive, vendetta-driven decisions targeting a single company — one whose principal offense, according to Congresswoman Zoe Lofgren of California, was refusing to surrender its technology to the administration’s preferred uses.
Get Involved Today
Contribute to our mission and turn your concerns into action.
What this timeline shows is not a coherent national-security strategy. It is a pattern of personalized, retaliatory action against a company that committed the original sin of refusing to grant unrestricted military and surveillance access. When the administration then turned around and used the very same technology for Iran strikes, no one in the West Wing seemed to notice — or care — about the contradiction.
“The Trump administration’s bullying tactics towards Anthropic are shocking and senseless.”
— Rep. Zoe Lofgren (D-CA), Ranking Member, House Science Committee
4. What This Tells Us About the President
A coherent national-security policy on frontier AI is possible. There are serious people in both parties who have written serious things about export controls, model evaluations, and the legitimate role of government in restricting the most dangerous capabilities. Anthropic itself, in a March 2025 filing with the Office of Science and Technology Policy, called for stronger export controls and a tougher China posture. The company is not opposed to oversight. It is opposed to oversight by ambush.
What we are watching is something else: a presidency in which complex technological, economic, and national-security questions are resolved by whether the President — or whichever adviser has his ear that morning — happens to be angry at a given company. The Friday afternoon ultimatum, with no written technical finding, no statutory process, and no internal coordination with the agencies that actually use the technology, is the signature of an executive branch that has stopped doing the work of government.
It is the same signature visible in the Easter Sunday Truth Social post threatening Iran’s bridges and power plants; in the President’s aggressive insults of Pope Leo XIV; in his sharing AI-generated images casting himself as Christ. And it is the signature that, in April, prompted Senator Chris Murphy of Connecticut to write that anyone in Trump’s Cabinet should spend Easter “calling constitutional lawyers about the 25th Amendment.”
The Anthropic episode does not stand alone. It is the technology-policy expression of the same pattern of impulsive, grievance-driven, process-free decision-making the country has now watched for eighteen months in domain after domain.
A mechanism designed for exactly this moment — and the honest barriers in its way.
Section 4 of the Twenty-Fifth Amendment to the Constitution was ratified in 1967 precisely to address a President “unable to discharge the powers and duties of his office.” Its drafters — chastened by the assassination of John F. Kennedy and the prolonged incapacitation of Woodrow Wilson — wrote a deliberately narrow mechanism: the Vice President and a majority of the Cabinet (or a body Congress designates) declare the President unable to serve, and the Vice President assumes the duties of acting President. If the President contests, a two-thirds vote of both chambers of Congress is required to keep him from resuming office.
The Amendment was not designed for partisan disputes. It was designed for the moment when the country must reckon honestly with whether the person holding the nuclear codes can still discharge the office’s duties. That moment is here, and named officials are saying so on the record.
Who Has Called for Action
On April 10, House Judiciary Ranking Member Jamie Raskin formally requested that White House Physician Captain Sean Barbabella conduct “a comprehensive neuropsychological assessment” of President Trump. On April 14, Raskin and fifty House Democratic co-sponsors introduced legislation establishing the bipartisan commission expressly contemplated by Section 4 of the Amendment. On April 30, Senators Sheldon Whitehouse and Jack Reed of Rhode Island entered into the Congressional Record a statement signed by thirty-six physicians — neurologists, psychiatrists, and cognitive specialists from Harvard, Tufts, Columbia, and George Washington University — describing the President’s “rapidly worsening, reality-untethered, increasingly dangerous decline” and calling for the Amendment’s use.
The Constitutional Argument
The argument is not that the President holds bad policies, or that he is impulsive, or that he is rude. The argument is more specific and more grave: that the cumulative pattern — middle-of-the-night threats to attack foreign infrastructure, vendettas against private companies dressed up as national-security policy, decision-making by ninety-minute ultimatum without written findings — is no longer consistent with the deliberative judgment the office requires. The Anthropic shutdown is not the worst example. It is simply a clear, dated, well-documented one.
The Honest Barriers
The barriers are real and must be named. Section 4 requires the Vice President’s participation, and Vice President Vance is unlikely to initiate it. The Cabinet is composed of officials selected for loyalty. If the President contests removal, two-thirds of both chambers of Congress must vote against him — a margin that does not currently exist. The constitutional path is, in practical terms, blocked.
Why the Blockage Does Not Settle the Matter
It is precisely because the Amendment’s political path is hard that its moral force matters. The framers wrote Section 4 to provide a record — a written, dated, public account that the country was warned by the people closest to the Constitution. That record is being built now, by Raskin’s commission bill, by the thirty-six physicians’ Congressional Record statement, by senators of both parties who privately concede what they will not yet say in public. The Anthropic episode is now part of that record. The fact that removal is hard is not an argument that the warning is wrong. It is an argument that the country must elect, in 2026 and 2028, the Congress and the Cabinet capable of acting on it.
5. What Must Happen Now
Three things are required, none of them partisan.
First, the export-control authority being used here must be brought under statutory discipline. If the Commerce Department can disable an American company’s flagship product on a Friday afternoon with no written technical finding, then there is no rule of law in American technology policy, only ruler-of-the-day discretion. Congress — including the Republicans who once cared about administrative due process — must require a written, evidence-based finding before any future order of this magnitude.
Second, the bipartisan commission contemplated by the Twenty-Fifth Amendment, and proposed in Raskin’s April 14 legislation, must receive a floor vote. Members of both parties should be required to state, on the record, whether they believe a sitting President’s fitness for office is something the Constitution permits Congress to examine. The Amendment is not a partisan tool. It is a written mechanism in the founding document. Refusing even to vote on a commission to invoke it is itself a constitutional failure.
Third, American voters must understand what they are watching. The Anthropic shutdown is a window into the decision-making style now operating at the highest level of American government. The pattern is consistent: grievance, retaliation, ultimatum. The same instincts that ordered a Friday afternoon kill switch on Anthropic are the instincts that posted “Power Plant Day, and Bridge Day” on Easter morning. They are not separate phenomena. They are the same President.
Editorial Conclusion
A nation governed by ninety-minute ultimatums is not, in any meaningful constitutional sense, governed at all. The Anthropic shutdown is not the largest scandal of this administration, and it will not be its last. But it is the clearest possible illustration of what happens when impulse replaces process and grievance replaces statute. The Twenty-Fifth Amendment is not a partisan weapon — it is a written, ratified mechanism placed in the Constitution by Americans who understood that the question of presidential capacity must never be unaskable. It is being asked now, by named lawmakers and named physicians, on the record. The country owes itself an honest answer.
Sources & References
- Washington Post — How Anthropic lost the White House’s trust — and then its flagship product
- Anthropic — Statement on the US government directive to suspend access to Fable 5 and Mythos 5
- Time — Anthropic Pulls Its Most Powerful AI Models After U.S. Bars Foreign Access
- Al Jazeera — US asks Anthropic to block global access to top AI models: Why it matters
- TechCrunch — Anthropic’s safety warnings may have just backfired
- Semafor — White House move to limit Anthropic linked to concerns about Chinese access to Mythos
- Fortune — A warning from Amazon led the White House to shut down Anthropic’s Mythos model
- Foreign Policy — The Anthropic-Trump Fight Over Claude Mythos 5 and Fable 5, Explained
- TheStreet — Anthropic heads to Washington to win its AI models back
- Yahoo Finance — Anthropic to meet with Trump administration over AI cybersecurity dispute
- The Next Web — Anthropic heads to Washington for crisis talks as Fable 5 ban spirals
- The Hacker News — U.S. Orders Anthropic to Suspend Fable 5 and Mythos 5 Access for Foreign Nationals
- Memeburn — Anthropic Forced to Shut Down Claude Fable 5 and Mythos 5 (Business Impact)
- Traders Union — Anthropic shutdown order hits pre-IPO share market
- House Science Committee Dems — Ranking Member Lofgren Appalled by Trump Administration’s Attacks on Anthropic
- House Judiciary Committee Dems — Raskin Demands Cognitive Evaluation of Trump, Calls to Invoke 25th Amendment
- Mediaite — House Democrats File Bill to Form 25th Amendment Commission
- The Hill — Trump’s Decline and the 25th Amendment — Physicians on the Record
Related News
