Anthropic, the San Francisco-based artificial intelligence company behind the Claude family of AI models, has released Fable 5 — a restricted but publicly accessible version of its most advanced AI system — to general users on Tuesday, 9 June 2026. The launch marks the first time technology from the company's Mythos class, its most capable AI lineup, has been made broadly available, following months of limited access due to serious cybersecurity concerns.
Fable 5 and its unrestricted counterpart, Claude Mythos 5, are essentially the same underlying model but differ in the safeguards applied and the audience permitted to use them. The full Mythos version remains accessible only to a select group of roughly 200 organisations across more than 15 countries enrolled in Anthropic's Project Glasswing programme — a cybersecurity partnership that has expanded since the model's April unveiling. Organisations in that programme have already used Mythos to identify more than 10,000 critical security vulnerabilities in real-world systems. The public-facing Fable 5, by contrast, routes sensitive queries related to cybersecurity, biology, and chemistry to a lower-tier model, Opus 4.8, released in late May. Anthropic says it hired outside experts to spend more than 1,000 hours attempting to bypass these restrictions — a process known as red-teaming — and ran a bug bounty programme; no one succeeded in fully unlocking the model.
The capabilities drawing both excitement and unease are significant. Anthropic describes Fable 5 as exceptionally strong at writing and debugging software code, answering complex research questions, and analysing images. Both Fable 5 and Mythos 5 can also operate "unattended" — carrying out tasks autonomously over longer periods than any previous Claude model — a feature that has heightened concerns about oversight. "Right now, it's like the AI industry has a gas pedal, but it doesn't have a brake pedal," one observer noted. Fable 5 carries a steep price: $10 per million input tokens and $50 per million output tokens, double the cost of Opus 4.8.
The release arrives amid a fraught political backdrop. Anthropic is engaged in an ongoing legal dispute with the US Department of Defense after refusing to lift restrictions on use cases including mass surveillance and autonomous lethal weapons — a standoff that led the Pentagon to sever contracts with the company. Nevertheless, US government agencies have tested Mythos, and the White House has established a framework for reviewing the most powerful AI models before public release. The Canadian Finance Minister earlier described the attention around Mythos as warranted, citing concern about "unknown unknowns."
Why this matters: The Fable 5 launch crystallises a tension at the heart of the AI industry — between commercial pressure to release ever-more-capable tools and genuine uncertainty about the risks they carry. Anthropic acknowledged plainly that "releasing a model this capable comes with risks." The company also disclosed it had detected large-scale attempts to extract its technology for training competing models in authoritarian states. With both Anthropic and rival OpenAI having recently filed IPO plans, and the broader AI sector drawing intense investor interest, the stakes of getting this balance right — commercially, strategically, and ethically — have rarely been higher.