Anthropic launches Claude Fable 5 AI model with safety guardrails

A recently released artificial intelligence tool, Claude Fable 5, from Anthropic, has raised concerns due to its power, previously deemed too significant for public release.

Initially made available for preview in April as part of Anthropic’s Claude Mythos, this AI program has stirred apprehension among leaders in technology, finance, and government, with worries about potential financial security risks.

While some view the warnings as exaggerated marketing, the company acknowledges the risks associated with releasing such a capable model, stating that its capabilities surpass those of any model they’ve previously made public.

As part of its launch, Fable includes user limitations and safeguards. Despite a lawsuit against the US Department of Defense regarding the use of its AI tools, US government agencies have begun testing Mythos.

Anthropic’s valuation approaches $1 trillion, highlighting its attractiveness to investors due to its superior AI capabilities.

The approximately 150 organizations that previewed Mythos will now access Claude Mythos 5, with specific cybersecurity and biological limitations tailored to organizational needs.

These users have already identified over 10,000 critical security flaws through the tool.

Anthropic plans to expand access to Mythos 5 through a broader trusted access program, allowing both Fable and Mythos to operate unattended for extended periods based on human commands—longer than previous models.

Co-founder Jack Clark emphasized the need for a balance in AI development, stating that while the industry accelerates rapidly, there should be mechanisms to decelerate its advancement for safety purposes.