U.S. Export-Control Order Forces Anthropic to Disable Fable 5 and Mythos 5 for All Customers
The Trump administration placed Anthropic’s Fable 5 and Mythos 5 under export controls, telling the company in a letter to Dario Amodei that foreign governments, companies and individuals could no longer access the models. Anthropic said the directive, which it says also applies to foreign nationals inside the US, including its own employees, meant it had to abruptly disable both models for all customers to ensure compliance. All other Claude models remain available; for Fable 5, new sessions will run on users’ default model or Opus 4.8, while existing sessions and API requests will return errors.
Anthropic said it believes the order is a misunderstanding and is working to restore access. In a later company response, it said the government had provided only verbal evidence of a potential narrow, non-universal jailbreak involving requests to read a specific codebase and fix software flaws, and argued the capability shown is widely available in other models including OpenAI’s GPT-5.5. The directive lands days after Anthropic launched Fable 5 as the public version of Mythos and after it walked back hidden limits on frontier-LLM-development requests; Arena said it has now removed Fable 5 after previously ranking it No. 1 across Agent, Text and Code.
From the sources (25 posts)
@fox0x01RT @logangraham: New post on Red today: Our team @AnthropicAI found that Mythos Preview is meaningfully better at developing N-days. It to…
@ethancaballeroRT @difficultyang: Fable's analysis on what the "limiter" might look like in claude code. But the sessions it is looking at are not finishe…
@andersonbcdefgRT @vikhyatk: Was using Fable 5 to write inference code Anthropic flagged it as frontier AI research steering vector kicked in and it sta…
@reutersAnthropic rolls out public version of Mythos without cybersecurity capability
@blancheminervaDoD: “Anthropic is a supply chain risk” Anthropic: “You ain’t seen nothing yet.”
@asvanevikfirst impression of Fable: engineermaxxed. clearly optimized for coding, to the point where it feels autistic on other things. on a few strategy / reasoning / writing tasks, the responses felt overcomplicated, too detailed, and missing t
@wesrothFable 5 reportedly leads on most tested AI capability benchmarks. It performs especially well in software engineering, knowledge work, vision, and scientific research. The most interesting claim is that it gets better as tasks get longer a
@jeremyphowardRT @natolambert: A message to Anthropic leadership: You're not special. Making sure AI goes well is a team effort not a "you effort."
@sbmarufRT @BlancheMinerva: DoD: “Anthropic is a supply chain risk” Anthropic: “You ain’t seen nothing yet.”
@tolyWell, what sucks is that basically I’ve had both 4.8 and 5.5 spinning over and over and probing percolator for bugs for the last week and basically nothing, and within the first hour fable found what seems like a credible LOF in the engine.
@theo@LowLevelTweets Mythos was not meant to be a cyber expert. That was an emergent behavior of being good at coding. Fable is as good as Mythos at most code things. It literally is just Mythos, but with filters before it to determine if your
@code_starRT @code_star: It’s almost certainly because they did some kind of A/B testing. My guess is they defined frontier models development so br…
@zooko@lacker 🤔 Is this Fable working around a limitation that you want it to have, or—more likely—you and Fable working around a limitation that Anthropic wants it to have? Or—most likely of all—you and Fable working around a limitation nobody w
@blackhcI'm confused why so many are upset and surprised that they can't use Claude anymore to catch up with Anthropic - or at least they won't get the same benefits (it is not clear how much worse the model is vs Mythos for frontier RE tasks) The
@lcastricatoHas anyone confirmed if fable is degraded solely on frontier llms or all frontier models? Does it need to be autoregressive?
@ethancaballerothe claude fable 5 nerf for AI research has induced the angriest reaction from AI researchers that I've ever seen in my life
@deredleritt3r@BlackHC I'm absolutely fine with Anthropic limiting these kinds of use cases. I don't understand why anyone would have expected the frontier labs to be fine with other people (including, potentially, their competitors) using their models
@lcastricatoRT @0xBalloonLover: anthropic won't let you use fable for biology, chemistry, ai research, or anything that accelerates human progress. tha…
@iscienceluvrRT @lcastricato: Has anyone confirmed if fable is degraded solely on frontier llms or all frontier models? Does it need to be autoregressiv…
@wesrothClaude Mythos 5 is being positioned as a major step forward for AI-assisted scientific research, especially in biology, drug design, protein engineering, and genomics. The examples show Mythos 5 doing more than answering questions. It can
@togeliusRT @natolambert: A message to Anthropic leadership: You're not special. Making sure AI goes well is a team effort not a "you effort."
@code_starRT @code_star: Makes me wonder how long this has already been going on without users being notified. I had been feeling like codex was ru…
@wightmanrRT @natolambert: A message to Anthropic leadership: You're not special. Making sure AI goes well is a team effort not a "you effort."
@reutersAnthropic is rolling out a public version of its Mythos AI model, but with guardrails barring its use in risky areas such as cybersecurity, after a preview earlier this year sent shockwaves globally with its ability to find software flaws h
@jeremyphoward@BlackHC It's very very very unsafe. Absolutely disastrously unsafe. You shouldn't be surprised that people are horrified about this.