Anthropic, the developer behind the Claude AI series, has officially released 'Claude Fable 5' to the public—a move that directly contradicts its own previous warnings that the model was too powerful for general release.
The company previously admitted that its Claude Mythos program was so intelligent it could potentially be weaponized to exploit or hack computer systems.
Despite these alarming risks, Anthropic is now pushing the technology into the wild, claiming that safeguards are in place while simultaneously acknowledging that 'releasing a model this capable comes with risks.' The company’s own co-founder, Jack Clark, recently admitted the industry is effectively driving without a brake pedal, noting that the rapid advancement of AI needs a mechanism to slow down.
Yet, instead of exercising caution, Anthropic is expanding access to these high-level tools, including versions that lack specific cybersecurity or biology limitations for select organizations.
While the company touts the discovery of 10,000 security flaws by early testers as a success, the decision to broaden access to such potent technology highlights the reckless pace of the AI arms race and the inability of developers to truly control the digital genie they have let out of the bottle.
Tags
