Anthropic has released Claude Fable 5 to the public after previously saying the underlying Claude Mythos model was too powerful for broad public access.
Claude Fable 5 is a version of Claude Mythos, the AI program that drew concern from technology, finance, and government leaders after a private preview in April, according to BBC Tech. The release turns Anthropic’s central safety dilemma into a public test: whether a model the company itself framed as risky can be made widely available without handing users dangerous capabilities.
Anthropic releases Claude Fable 5 after earlier concerns over public access
Anthropic said Tuesday that Fable 5 will be released with safeguards and user limitations. It did not present the launch as an unrestricted release. The company also acknowledged the risk directly.
"releasing a model this capable comes with risks"
Anthropic added:
"Fable's capabilities exceed those of any model we've ever made generally available"
That wording matters. Anthropic is not saying Fable 5 is merely another model update. It is saying this is the strongest Claude model it has put into general availability, even as it keeps some controls in place.
The public version is tied to Claude Mythos, which Anthropic made available privately in April to a small group of organizations for previewing and testing. At the time, Anthropic said Mythos was intelligent enough to be dangerous because of its ability to exploit or hack computer systems.
The company also said about 150 groups that had preview access to Mythos will now get access to Claude Mythos 5. That version does not have limitations on cybersecurity or biology, depending on the organization’s specific uses.
The distinction is sharp:
| Model | Access | Limits described by Anthropic | Core issue |
|---|---|---|---|
| Claude Fable 5 | Public release | Safeguards and user limitations | Strongest Claude model generally available |
| Claude Mythos 5 | Roughly 150 groups, expanding later | No limitations on cybersecurity or biology depending on use | Higher-risk access for selected organizations |
Anthropic said the updated Mythos access is currently limited to a "small group of cyberdefenders and infrastructure providers", but it expects that group to grow.
"We intend to expand access to Mythos 5 through a broader trusted access program"
Claude Mythos controversy puts Anthropic’s safety claims under pressure
The controversy around Claude Mythos started with capability, not branding. Anthropic said the model could pose danger because of its ability to exploit or hack computer systems. That made the private April preview unusually sensitive.
Canadian Finance Minister François-Philippe Champagne told the BBC in April that attention on Mythos was warranted in part because:
"it's the unknown, unknown"
That phrase captures the policy problem. The model’s risks are not limited to known misuse cases. The concern is that highly capable AI systems may reveal new attack paths after release, especially when they can operate for longer without direct human supervision.
Anthropic said both Fable and Mythos can work "unattended" on human commands for longer periods "than any previous Claude models." That is the technical feature behind much of the concern. Longer unattended operation can make an AI tool more useful for legitimate tasks, but it also raises the stakes for monitoring, permissioning and audit trails.
The company has also reported a concrete defensive benefit. Groups and companies using Mythos have reported finding more than 10,000 critical security flaws in their systems, according to Anthropic.
That creates the central tension. The same capability that helps cyberdefenders find flaws may also be the capability that makes policymakers uneasy about broader access.
MLXIO analysis: Anthropic’s safety reputation will now be judged less by its stated caution and more by how well the Fable 5 limits hold up in public use. A controlled release is still a release. If users find ways around safeguards, the company’s earlier warnings about Mythos will become evidence against it.
Public access could increase pressure on AI companies and regulators
Anthropic’s release lands as the company is expected to become a public company soon, with its private valuation nearing $1tn (£747bn), according to the BBC report. Better and broader AI capabilities support that investor narrative. They also raise the cost of restraint.
The company is already entangled with government use. The BBC reported that Anthropic has an ongoing lawsuit against the US Department of Defense over its refusal to accept government use of its AI tools. At the same time, US government agencies have been testing Mythos.
That contradiction is now harder to ignore. Anthropic is restricting, litigating and expanding access at once. Each move may be defensible on its own terms, but together they show how difficult it is to draw clean lines around frontier AI use.
For broader MLXIO context on how leaders often misread emerging technology risk, see Future Trends Everyone Keeps Misreading — Here's Why. The Fable 5 release fits that pattern: the real issue is not whether the model is impressive, but whether institutions can assess it fast enough.
A second MLXIO lens is policy pressure. Our coverage of Stake Grab Brings AI Companies to Trump's White House tracks how AI company power is becoming a political question, not just a product question. Fable 5 gives that debate a new test case.
MLXIO analysis: If Fable 5 performs close to Mythos while carrying public-facing safeguards, enterprise customers and public agencies may want access to stronger versions under negotiated controls. That would push Anthropic toward more tiered access programs, not fewer.
Anthropic still has to prove the safeguards match the model
The immediate unanswered questions are practical. Anthropic has said Fable 5 includes safeguards and user limitations, but the BBC report does not detail exactly what those safeguards block, how usage limits work, or how sensitive cybersecurity and biology capabilities are restricted in the public version.
The technical gap between Claude Fable 5 and Claude Mythos 5 also remains unclear. Anthropic says they are essentially the same model with different safeguards and levels of access. That leaves users, companies and regulators watching for evidence in reasoning, coding, tool use, autonomy and security-related behavior.
The next signals will come from Anthropic’s documentation, early user reports, enterprise adoption and any incidents tied to public access. Regulator responses will matter too, especially because finance and government leaders were already paying attention before this release.
The practical watch item is narrow but consequential: whether Claude Fable 5 behaves like a carefully constrained public product, or whether its release shows that the frontier AI race is moving faster than the systems meant to contain it.
Impact Analysis
- Anthropic is publicly releasing a model it previously framed as too powerful for broad access.
- The launch tests whether safeguards can contain advanced AI capabilities without blocking useful applications.
- Selective access to less-limited Mythos 5 raises questions about who gets powerful AI tools and under what oversight.
