• Early warnings and controlled deployment
• Amazon’s concerns and a White House scramble
• The export‑control order and global shutdown
• Anthropic’s rebuttal
• Government and industry reactions
• Wider implications

US Government Orders Anthropic to Shut Down Fable 5 and Mythos 5 Models The sudden shutdown of Anthropic’s most powerful AI models, Fable 5 and Mythos 5, has become a test case for how far governments will go to treat frontier AI as a national security asset, and how hard industry will push back.

§Early warnings and controlled deployment

Anthropic’s CEO Dario Amodei had previously warned that the Mythos line was so powerful it posed “serious threats,” prompting a tightly controlled program in which Mythos 5 was shared with about 50 vetted organizations for defensive cybersecurity work rather than released broadly. Fable 5, a more heavily guarded, general-use variant of Mythos, launched publicly just three days before the shutdown and was quickly billed as Anthropic’s most capable model for everyday users.

§Amazon’s concerns and a White House scramble

Behind the scenes, Amazon security researchers assembled a report claiming they could prompt Fable 5 to surface information useful for cyberattacks. On Thursday evening and Friday morning, Amazon CEO Andy Jassy and at least five other companies flagged these findings to senior U.S. officials, triggering an urgent review inside the White House.

Administration officials, already wary after another company said it had “jailbroken” Mythos, pressed Anthropic for a voluntary pause but failed to secure one.

§The export‑control order and global shutdown

At 5:21 p.m. ET Friday, Commerce Secretary Howard Lutnick sent Anthropic an export-control directive barring use of Fable 5 and Mythos 5 by any foreign national, including foreign-born Anthropic staff. The order effectively treated the models as dual-use security technologies and placed Anthropic in a licensing regime that could impose penalties for noncompliance.

Because Anthropic could not easily segregate foreign users, the “net effect” was a worldwide blackout of both models by Friday night.

§Anthropic’s rebuttal

In a public statement, Anthropic said the government had provided only “verbal evidence of a potential narrow, non-universal jailbreak” that let Fable 5 review a specific codebase for “relatively simple” vulnerabilities that other public models, such as GPT‑5.5, could also find. The company insisted “this is a misunderstanding” and argued that treating such a narrow issue as grounds for recall would “essentially halt all new model deployments for all frontier model providers.”

Anthropic further said it had not received “a disclosure of a concerning non-universal potential jailbreak that led to a harmful result” and stressed that extensive red-teaming with U.S. and U.K. authorities had found its safeguards “substantially more effective” than previous models.

§Government and industry reactions

Administration officials countered that until the “national security apparatus is hardened,” the models must stay locked down, potentially for several weeks. The Pentagon’s chief information officer publicly backed the move, saying “some things are simply more important than revenue cycles, clickbait, and pre-IPO valuation.”

Separately, the episode deepened an already fractious relationship between Anthropic and the Trump administration, which earlier labeled the company a supply-chain risk after disputes over military use and surveillance, and then moved to block federal agencies from using its AI.

§Wider implications

Analysts say the Fable/Mythos shutdown illustrates a reactive, security-first posture toward fast-moving AI—and highlights how a single corporate research memo can cascade into sweeping export controls affecting “hundreds of millions” of users worldwide.

Continue reading https://foxvector.com/stories/019ec48e-8fea-3e84-72c2-31e7fadc7a6e