A version of an artificial intelligence tool deemed too powerful for the public has been released to the general public. Anthropic on Tuesday made Fable 5 available – the first widely accessible model from its advanced Mythos class, which was restricted to a small set of partner institutions for months over cybersecurity concerns.
Fable 5 is a stripped-back version of Claude Mythos, an AI program that caused serious concerns among technology, finance and government leaders when it was previewed privately in April. Some worry the tool is so powerful it could pose financial security risks, though others have questioned how much of the hype is marketing spin.
“Anthropic made Fable 5 public after months of restricting its advanced Mythos AI over cybersecurity fears.”
Anthropic said Fable 5 would be released with safeguards and user limitations, but acknowledged that “releasing a model this capable comes with risks”. Most queries about cybersecurity, biology or chemistry will be routed instead to a less powerful model, Opus 4.8. The company also said it had identified large-scale attempts to extract its technology to train competing AI models in authoritarian countries, and those queries will also fall back to a weaker system.
When Mythos was first released to a small group of organisations, Anthropic said it was doing so because the tool was so intelligent it could be dangerous, due to its ability to exploit or hack computer systems. Canadian finance minister François-Philippe Champagne told the BBC in April that the attention on Mythos was warranted, in part, because “it’s the unknown, unknown”.
Despite an ongoing lawsuit between Anthropic and the US Department of Defense over government use of its AI tools, US government agencies have been testing Mythos. The White House has since set up an arrangement to test the most powerful models from leading AI companies before release.
Anthropic co-founder Jack Clark told BBC Newsnight that the ability of AI tools was expanding so rapidly that the company thought there should be a way for the public to slow the technology’s advancement. “You want the option to be able to take your foot off the gas and put your foot on the brake,” he said.
Alongside the public release, the roughly 150 groups that had been given access to preview Mythos will now have access to Claude Mythos 5, which does not have limitations on cybersecurity or biology, depending on an organisation’s specific uses. That group has been expanded to about 200 organisations in more than 15 countries and is expected to grow further. To date, groups and companies using Mythos have reported finding more than 10,000 critical security flaws in their systems.
Anthropic also said it had hired outside experts to spend more than 1,000 hours trying to find ways to bypass its restrictions – a process known as red-teaming. The company ran a bug bounty programme to pay people to find vulnerabilities.
Anthropic is expected to become a public company soon, with its private valuation nearing $1tn. Consistently showing better and broader AI capabilities underlines its value to investors.