Anthropic Releases Fable 5 After Calling It Too
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- Anthropic released Claude Fable 5 publicly, stating its 'capabilities exceed those of any model we've ever made generally available' while acknowledging 'releasing a model this capable comes with risks'
- Claude Mythos, released privately in April to a small group of organizations, caused serious concerns among technology, finance, and government leaders who feared it could exploit or hack computer systems
- François-Philippe Champagne, Canada's Finance Minister, told the BBC the attention was warranted because 'it's the unknown, unknown'
- Mythos users have reported finding more than 10,000 critical security flaws in their systems, according to Anthropic
- Anthropic is granting the roughly 150 groups that previewed Mythos access to Claude Mythos 5, which drops limitations on cybersecurity and biology depending on the organization's use case
- Jack Clark, Anthropic co-founder, told BBC Newsnight the AI industry 'has a gas pedal, but it doesn't have a brake pedal' and called for a public mechanism to slow AI advancement
- Anthropic's private valuation has neared $1tn (£747bn) as the company prepares to go public, with consistent capability demonstrations central to its investor appeal
Why it matters: Anthropic simultaneously told the public Fable 5 exceeds any model it has ever released while also saying the release 'comes with risks' — and Mythos users have already surfaced more than 10,000 critical security flaws. The contradiction is sharpened by the company's near-$1tn private valuation and expected IPO, which give it a direct financial incentive to showcase ever-more-capable models regardless of the safety concerns its own co-founder is raising.
