When the federal government quietly forced Anthropic to shut down its most advanced public AI models overnight, it showed how little control ordinary Americans have over the technologies reshaping their lives.
Story Snapshot
- The Trump administration ordered Anthropic to block its top AI models from all foreign nationals worldwide, citing national security.[2][3]
- To obey the order, Anthropic disabled Fable 5 and Mythos 5 for every customer, including Americans and the U.S. government itself.[1][2]
- The government has not publicly shown proof of serious harm, but acted on an alleged “jailbreak” that Anthropic calls minor.[2]
- The clash highlights a deeper fear across left and right: powerful officials and tech elites are making high‑impact decisions in the dark, with almost no public say.[2][5]
What Exactly Did The Government Do To Anthropic’s AI?
The United States government used its national security powers to issue an export control order against Anthropic’s newest artificial intelligence models, Fable 5 and Mythos 5.[2][3] The directive bars any foreign national from accessing these models anywhere in the world, even if they work for Anthropic or live inside the United States.[2][5] Because Anthropic cannot reliably separate foreign and non‑foreign users at scale, the “net effect” was that the company had to disable both models for every customer, public and private.[1][2]
Anthropic confirmed the order in an official public statement and in posts on social media.[1][2] The company said it received the directive in the early evening and had to cut off access abruptly to stay within the law.[2] Other Claude models, like the widely used non‑Mythos systems, remain online and were not included in this government action.[2][3] That split makes clear that Washington targeted only the very highest‑end tools, the ones the government itself had just begun testing for sensitive work.[6]
Why Did Washington Say These AI Models Were A Threat?
Federal officials have not released a detailed public explanation, but the order cites national security and export control laws focused on advanced technology.[2][3][5] According to Anthropic, the government believes someone found a way to “jailbreak” Fable 5, meaning trick the system into doing something it was designed not to do.[2] The specific claim is that the jailbreak could help find software vulnerabilities, which might matter for cyber defense or cyber attack, depending on who uses it.[2]
Anthropic says it reviewed a demonstration of the alleged jailbreak and found that it uncovered only a small set of minor security flaws that were already known from other sources.[2] The company also states that other public AI models can locate similar weaknesses without any special bypass tricks.[2] Anthropic notes it has not seen a “universal jailbreak” that lets users broadly escape the model’s safety rules, and it says nobody has shown it a case where this bug led to serious harm in the real world.[2]
Why Did Anthropic Shut Down Access For Americans Too?
Anthropic argues that the structure of the export order left it no real choice.[1][2] The directive bans foreign‑national access anywhere, including inside American borders and inside Anthropic’s own workforce.[2][5] In practice, that would require the company to track citizenship and immigration status across every account, every organization, and every shared environment where the models run. Anthropic says the only way to be sure it was obeying the law was to disable Fable 5 and Mythos 5 completely for all customers.[1][2]
This move also affects parts of the United States government that had been testing Mythos 5 through a pilot program known as Project Glasswing.[6] That means the same government that claimed to be acting to protect national security has also cut off its own access to a tool it considered useful enough to test for advanced work.[6] For many Americans, this looks like the kind of blunt, top‑down decision that fuels distrust across the political spectrum.
What This Fight Reveals About Power, Secrecy, And “The System”
This episode pulls together many frustrations shared by older conservatives and liberals alike. Conservatives see another case where unelected security officials and big tech firms make sweeping calls with limited transparency, while everyday Americans and small businesses just lose access to a powerful tool they were promised.[3][5] Liberals see a closed‑door process that could shape jobs, speech, and surveillance, with almost no public debate or protection for vulnerable groups who may be most affected by rapid automation.
Three days.
That's how long Anthropic's Fable 5 was publicly available before the US government pulled the plug – globally, for every customer, including inside the US.
It launched June 9 as their most capable public model (Mythos class). Yesterday they received an… pic.twitter.com/S7tNDPwQBA
— Oscar (@OscarActual) June 13, 2026
Across both sides, the pattern is familiar: insiders in Washington and Silicon Valley act first, explain later, and rarely face real accountability when they get it wrong.[2][3][5] The government says it is acting “out of an abundance of caution,” but it has not shared enough evidence for citizens to judge whether this shutdown was wise safety policy or clumsy overreach.[2] The deeper worry is that as artificial intelligence grows more powerful, these quiet directives will decide who gets to use that power, and for what purpose, long before voters ever get a say.
Sources:
[1] YouTube – Anthropic disables top AI models after US foreign access order
[2] YouTube – Anthropic Just Dropped Fable 5 And It’s Terrifying
[3] YouTube – Anthropic Just Dropped Claude Mythos and Fable 5 (Full Breakdown)
[5] YouTube – The US Government Just Shut Down Fable 5 + Mythos (Unbelievable)
[6] Web – Anthropic Fable 5, Mythos 5 disabled after US export control order