Anthropic Releases Limited Version of Its Most Advanced AI Model That Stoked Cyber Concerns

Artificial intelligence giant Anthropic released a “Mythos-class” model to the public on June 9, two months after the company rolled out the Mythos model only to a limited number of government and industry users over concerns it could be weaponized by the wrong people.

In a blog post on Tuesday, Anthropic said it was launching Claude Fable 5, a model like Mythos that the company said it “made safe for general use.”

“Fable 5’s capabilities exceed those of any model we’ve ever made generally available. It is state-of-the-art on nearly all tested benchmarks of AI capability, showing exceptional performance in software engineering, knowledge work, vision, scientific research, and many other areas,” Anthropic wrote.

“The longer and more complex the task, the larger Fable 5’s lead over our other models.”

In the announcement, the company said it was also launching Claude Mythos 5 to a confined group of cyberdefenders and infrastructure providers. Mythos 5 is the “same underlying model” as Fable 5, but with some safeguards lifted, Anthropic noted.

Last week, Anthropic called for industry leaders worldwide to slow down the development of AI, warning that the models are reaching a point at which they could soon improve themselves independently.

Some fear that if AI develops and improves itself without confinement, humanity could lose control of the technology.

“This is called recursive self-improvement,” Anthropic’s Marina Favaro and Jack Clark wrote in a blog post. “We are not there yet, and recursive self-improvement is not inevitable. But it could come sooner than most institutions are prepared for.”

In April, Anthropic warned of a potential cybersecurity reckoning from frontier AI models when first announcing a private rollout of Mythos Preview.

In its blog post on Tuesday, Anthropic said releasing a model as capable as Fable 5 “comes with risks,” and that with the safeguards removed, the model “could be misused to cause serious damage.”

During the April announcement of Mythos, Anthropic said the model was capable of finding “zero-day” exploits in critical computer software, including operating systems. Exploits are code that can take advantage of software vulnerabilities, and “zero-day exploits” are unknown to developers while designing software systems.

Anthropic explained that when users submit safety-sensitive queries to Fable 5, they will instead receive responses from the company’s “next-most-capable model, Claude Opus 4.8.”

“The capabilities of models like Fable 5 and Mythos 5 have the potential to do profound good for the world. We’ve seen the beginnings of this in Project Glasswing, where the models have helped cyber defenders secure critically important software,” Anthropic wrote.

“We’ve also seen it in life sciences research, where the models are positing novel hypotheses and speeding up the development of new therapeutics.”

Anthropic is also trying to go public. The company announced on June 1 that it had confidentially filed an initial public offering prospectus with the Securities and Exchange Commission.