Anthropic has introduced a powerful new artificial intelligence model, known as “Mythos,” as part of a broader push into cybersecurity, but the company is withholding a full public release amid concerns over the system’s potential misuse.
The mode, formally referred to as Claude Mythos Preview, is designed to autonomously detect and analyse software vulnerabilities across complex systems. Early tests show it can identify critical flaws in major operating systems and web browsers, including long-hidden bugs that had evaded human experts for years.
Anthropic said the system represents a significant leap in capability, particularly in its ability to uncover “high-severity vulnerabilities,” prompting the company to limit access to it. “Claude Mythos Preview’s large increase in capabilities has led us to decide not to make it generally available,” the company said in a statement.
Instead, the model is being deployed through a controlled cybersecurity initiative involving a select group of partners, including major technology firms and security organisations. More than 40 institutions are expected to use the system to scan their own infrastructure and open-source codebases for vulnerabilities.
Company executives emphasised that Mythos operates at a level comparable to highly skilled security researchers. Logan Graham, head of Anthropic’s frontier red team, described the model as “extremely autonomous,” with capabilities that rival those of advanced human experts in identifying and exploiting software weaknesses.
The system has already demonstrated its power by uncovering vulnerabilities across “every major operating system and web browser,” including some believed to be decades old.
However, these capabilities have also raised alarm. During internal testing, Mythos reportedly showed the ability to generate working exploits with minimal guidance and even bypass containment mechanisms, highlighting the risks of deploying such advanced tools broadly.
To address these concerns, Anthropic has launched a coordinated effort involving leading technology firms and security stakeholders to ensure the model is used primarily for defensive purposes. The initiative aims to strengthen global cybersecurity by enabling faster detection and remediation of critical vulnerabilities while developing safeguards against malicious use.
Industry analysts say the move signals a shift in how frontier AI systems are deployed, with companies increasingly prioritising controlled rollouts in sensitive domains such as cybersecurity. Anthropic indicated that broader access to Mythos-class systems will depend on the development of robust safety frameworks and governance mechanisms.
The announcement underscores both the promise and the risks of next-generation AI, as firms race to harness powerful models that can defend digital infrastructure, but could also be weaponised if left unchecked
Consider Reading:
- Cybersecurity Key to National Growth and Digital Development
- Glemad Launches AI System to Counter Cyber Threats Across Africa
Senior Reporter/Editor
Bio: Ugochukwu is a freelance journalist and Editor at AIbase.ng, with a strong professional focus on investigative reporting. He holds a degree in Mass Communication and brings extensive experience in news gathering, reporting, and editorial writing. With over a decade of active engagement across diverse news outlets, he contributes in-depth analytical, practical, and expository articles exploring artificial intelligence and its real-world impact. His seasoned newsroom experience and well-established information networks provide AIbase.ng with credible, timely, and high-quality coverage of emerging AI developments.