Anthropic releases public version of AI model once deemed ‘too powerful’

SAN FRANCISCO, USA — Artificial intelligence company Anthropic has released a public version of an advanced AI model that it previously restricted due to concerns over its powerful capabilities and potential security risks.

The newly launched Claude Fable 5 is a consumer-accessible version of Anthropic’s experimental Claude Mythos model, which was first made available in April to a limited group of organizations for testing and evaluation.

When Mythos was initially introduced, Anthropic said the system’s advanced reasoning and cybersecurity capabilities raised concerns about potential misuse, including the possibility of exploiting vulnerabilities in computer systems. As a result, access was tightly controlled and limited to select partners.

Despite those concerns, Anthropic announced on Tuesday that Claude Fable 5 would now be available to the public with built-in safeguards, usage restrictions, and monitoring systems designed to reduce potential risks.

“Fable’s capabilities exceed those of any model we’ve ever made generally available,” the company said, while acknowledging that releasing such a powerful model comes with challenges.

According to Anthropic, both Fable and Mythos are based on the same underlying AI system but differ in the level of access and safety controls applied. The models are capable of carrying out complex tasks for extended periods with minimal human supervision, a capability the company says surpasses previous Claude releases.

Anthropic reported that organizations testing Mythos have already identified more than 10,000 critical security vulnerabilities within their systems, highlighting the model’s potential value for cybersecurity and infrastructure protection.

The company said around 150 organizations that participated in the preview program will continue to have access to Claude Mythos 5, a version with fewer restrictions tailored for specialized cybersecurity and research applications.

The release comes amid growing debate over the rapid advancement of artificial intelligence. While supporters argue that increasingly capable AI systems can improve productivity, security, and scientific research, critics warn that the technology may outpace existing safeguards and regulatory frameworks.

Anthropic co-founder Jack Clark recently emphasized the need for mechanisms that would allow society to slow AI development if necessary.

“You want the option to be able to take your foot off the gas and put your foot on the brake,” Clark said, comparing the current pace of AI innovation to a vehicle accelerating without an effective braking system.

The launch of Claude Fable 5 is expected to intensify competition in the AI industry as companies race to develop more capable models while balancing concerns surrounding security, safety, and responsible deployment.