Key takeaways:
- Anthropic said Claude Fable 5 is its most capable generally available AI model and belongs to the same technology family as Claude Mythos.
- Fable 5 will restrict many prompts involving cybersecurity, biology and chemistry and route some users to the older Opus 4.8 model.
- About 150 organizations that had access to Mythos Preview will now receive Claude Mythos 5, with Anthropic saying users have reported more than 10,000 critical security flaws found in their systems.
Anthropic released Claude Fable 5 to the public Tuesday, giving users access to a version of the company’s most powerful class of artificial intelligence systems while imposing safeguards on areas it says could be dangerous.
The company said Fable 5 is the first generally available product in the same family as its Claude Mythos models, which drew attention earlier this year for advanced cyber capabilities. Mythos Preview, released privately in April to selected organizations, was described by Anthropic as powerful enough to find and exploit computer system vulnerabilities. According to NBC News, the model found thousands of critical and severe vulnerabilities, including bugs and exploits in major operating systems and web browsers.
“Fable’s capabilities exceed those of any model we’ve ever made generally available,” Anthropic said in a blog post. “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 said Fable 5 uses the same tier of technology as Mythos but is being released with guardrails and user limits. The company acknowledged the risks of making such a system widely available. “The same queries that are beneficial in the hands of cybersecurity professionals and biology researchers could be dangerous if available to malicious actors,” it wrote.
To reduce those risks, Anthropic said Fable 5 will block many responses to prompts involving sensitive areas such as cybersecurity, biology and chemistry. In those cases, it will route users to an earlier, less powerful model called Opus 4.8, which Anthropic said had been the highest-performing publicly available model until Tuesday.
“Fable 5’s capabilities in areas like cybersecurity, biology and chemistry are advanced enough that we’re taking a deliberately conservative approach for these topics at launch,” Anthropic told NBC News. “To enable general availability of other Mythos level capabilities, we’ve decided to deploy safeguards that err on the side of caution, applying broad restrictions to these topics for now.”
The BBC reported that some senior figures had worried Mythos could pose financial security risks, while others questioned whether some of the attention reflected marketing hype. Canadian Finance Minister François-Philippe Champagne told the BBC in April that the concern was warranted in part because “it’s the unknown, unknown.”
Anthropic said Fable 5 will be available at no additional cost to users on its Pro, Max, Team and seat-based Enterprise plans until June 22. After that, users will need to buy and use extra computing credits to access it. The company said it aims to add Fable 5 to its regular subscription plans “as quickly as we can,” though NBC News reported that Anthropic has struggled to meet rising demand for its AI systems.
The company also said the roughly 150 organizations that previously had access to Mythos Preview will now receive access to an upgraded model, Claude Mythos 5. Anthropic said those organizations included financial institutions, software companies and healthcare networks using the system to strengthen cyber defenses. The BBC reported that groups and companies using Mythos have told Anthropic they found more than 10,000 critical security flaws in their systems.
Anthropic said Mythos 5 broke performance records in several fields, including drug design and molecular biology, and described it as the company’s “first model to consistently produce novel, compelling scientific hypotheses.” The company said the updated Mythos access is limited to a “small group of cyberdefenders and infrastructure providers” but added, “We intend to expand access to Mythos 5 through a broader trusted access program.”
Both Fable and Mythos can work “unattended” on human commands for longer periods than previous Claude models, Anthropic said, according to the BBC.
The release came one week after President Donald Trump signed an executive order creating a voluntary process for AI companies to share systems with the government for safety testing before public release and to strengthen government cyber defenses. Anthropic told NBC News it has offered the federal government early access to all its models for years.
Be First to Comment