‘Safe’ version of Anthropic’s Mythos model hits market
Anthropic officially released Claude Fable 5 on Wednesday, allowing the public to access the core architecture of its notorious ‘Mythos’ model for the first time – a technology previously deemed to dangerous for distribution.
Fable 5 is being rolled out via Anthropic’s interace, but the launch comes with strict limitations.
The San Francisco-based AI titan confirmed it will block or defer queries in high-risk zones, including cybersecurity or chemistry, acknowledging that “releasing a model this capable comes with risks”.
To combat jailbreaks, Anthropic has also mandated a 30-day data-retention policy for all Fable 5 and Mythos 5 traffic.
The underlying Mythos engine shocked the tech and finance sectors, as well as public bodies, after it restricted launch in April.
Developed under the White House’s watchful eye and restricted to a handful of entities like Amazon and Microsoft, Mythos was designed to map realistic attack paths across critical infrastructure.
Anthropic previously admitted the system had already “found thousands of high-severity vulnerabilities, including some in every major operating system and web browser”, with some legacy bugs dating back 25 years.
The firm warned that the system could outperform “all but the most skilled humans” at exploiting software flaws, or spark a devastating wave of automated cyberattacks.
Guardrails or an admission of danger?
The decision to release a moderated version via Fable 5 has revived a conversation among security leaders about whether interface-level restrictions can truly contain offensive-grade AI.
Andrew Rubin, founder and chief executive of Illumio, said: “The introduction of guardrails isn’t evidence that the problem is solved – it’s an admission that even the companies building these models don’t fully trust where the capability leads.”
“Constraints at the interface don’t change the underlying math. Attackers won’t operate at that layer. They’ll go straight after the capability itself. And as these tools become more broadly available, the speed and scale of attacks will only increase.”
Some also see the simultaneous deployment of Fable 5 and the highly restricted Mythos 5 reveals Anthropic’s dual-track commercial strategy.
Douglas McKee, director of vulnerability intelligence at Rapid7, said: “Fable 5 is the everyday workhorse. Mythos 5 is the power tool for tightly vetted security and infrastructure teams.”
As Anthropic edges closer to a highly anticipated IPO, the commercialisation of Mythos-class capabilities is forcing businesses to rethink their security budgets.
However, as Roman Stanek, founder and chief executive at Gooddata, said: “Open source security, legacy code debt, infrastructure hygiene, they are all solvable problems, but all chronically underfunded. Now we have Mythos, this new shinier, more expensive model that can theoretically find and fix them.”
“Great. Except the incentive structure hasn’t changed. Nobody wanted to pay a human engineer to fix it. They’re not going to pay an AI to fix it either. The issue has never been capability, people are just unwilling to invest and until that changes the vulnerabilities will stay as the threat evolves.”