SAN FRANCISCO, UNITED STATES (Web Desk) – Artificial intelligence company Anthropic has released a public version of an advanced AI model that it previously considered too powerful for widespread use, marking a significant development in the rapidly evolving AI industry.
The new model, named Claude Fable 5, is based on Anthropic’s highly capable Claude Mythos system, which was initially made available only to a limited number of organisations for testing and evaluation. The company had restricted access to the original model due to concerns that its advanced capabilities could potentially be misused for cybersecurity attacks or other high-risk activities.
Announcing the launch, Anthropic acknowledged that the public release comes with inherent risks but said a range of safeguards, usage limits and monitoring systems have been implemented to reduce the potential for misuse.
“Fable’s capabilities exceed those of any model we have previously made generally available,” the company said, while stressing that responsible deployment remains a priority.
When Claude Mythos was first unveiled in April, the model attracted significant attention from technology experts, financial institutions and government officials. Anthropic had warned at the time that the system possessed unprecedented reasoning and problem-solving abilities, raising concerns about its potential to identify vulnerabilities in computer systems or conduct sophisticated cyber operations.
Despite those concerns, supporters argued that the technology could provide major benefits in cybersecurity, scientific research and complex business applications if used responsibly.
According to Anthropic, approximately 150 organisations were granted early access to Mythos for testing purposes. The company reported that users of the system identified more than 10,000 critical security vulnerabilities across various digital infrastructures, demonstrating both the model’s capabilities and its potential value for cyber defence.
Anthropic said some trusted organisations will continue to receive broader access to Claude Mythos 5, including fewer restrictions in specialised areas such as cybersecurity and biological research, depending on their authorised use cases.
The company added that access is currently limited to selected cyber-defence groups and infrastructure providers but is expected to expand through a broader trusted-access programme in the future.
A key feature of both Claude Fable 5 and Claude Mythos 5 is their ability to operate on assigned tasks for extended periods with minimal human supervision. Anthropic claims the models can independently carry out complex instructions for longer durations than any previous Claude system, allowing them to tackle sophisticated workflows and research projects.
The release comes as competition among leading AI developers intensifies. Anthropic has emerged as one of the industry’s most valuable firms and is reportedly approaching a valuation of nearly $1 trillion, with investors closely watching advancements in its technology portfolio.
Meanwhile, debates over AI safety and regulation continue to grow. Anthropic co-founder Jack Clark recently argued that society should have mechanisms to slow the pace of AI development if necessary, warning that technological progress is accelerating faster than existing safeguards.
“You want the option to take your foot off the gas and put your foot on the brake,” Clark said, emphasising the need for stronger oversight and public accountability as increasingly powerful AI systems enter the market.
The public release of Claude Fable 5 could accelerate AI adoption across industries while intensifying discussions around cybersecurity, regulation and the responsible use of advanced artificial intelligence.
