In a cautious move, Anthropic has delayed the public release of its most powerful AI model, Claude Mythos, citing serious concerns around cybersecurity and potential misuse.
The company stated that the model is highly capable of discovering and exploiting software vulnerabilities, raising risks for economies, public safety, and national security. Instead of a general rollout, Mythos is being deployed to over 40 organizations to help scan and secure their own code and open-source systems.
Highlighting the growing risks of advanced AI, Anthropic said in a blog post, “AI models have reached a level of coding capability where they can surpass all but the most skilled humans at finding and exploiting software vulnerabilities.”
In a related video, CEO Dario Amodei said, “Capabilities in a model like this could do harm if in the wrong hands, and so we won’t be releasing this model widely. More powerful models are going to come from us and from others, and so we do need a plan to respond to this.”
Anthropic revealed that Claude Mythos has already identified thousands of zero-day vulnerabilities across major operating systems and web browsers. In internal benchmarks, Mythos Preview scored 83.1% in tasks measuring its ability to recreate known security flaws and generate exploit code.
The model also uncovered a 27-year-old vulnerability in OpenBSD that could allow remote crashes, and identified multiple vulnerabilities in the Linux kernel, chaining them together to gain full system control.
Logan Graham, Anthropic’s Head of Frontier Red Teams, described the model as “extremely autonomous” with advanced reasoning capabilities, capable of identifying “tens of thousands of vulnerabilities” and even writing exploits—tasks that would challenge even top security researchers.
To address these risks, Anthropic has launched Project Glasswing, involving 12 major organizations including Amazon Web Services, Apple, Broadcom, Cisco, CrowdStrike, Google, JPMorgan Chase, the Linux Foundation, Microsoft, Nvidia, and Palo Alto Networks.
The company has committed $100 million in model usage credits for participants and donated $2.5 million to the Linux Foundation and $1.5 million to the Apache Software Foundation to strengthen open-source security against AI-driven threats.
Also read: Viksit Workforce for a Viksit Bharat
Do Follow: The Mainstream LinkedIn | The Mainstream Facebook | The Mainstream Youtube | The Mainstream Twitter
About us:
The Mainstream is a premier platform delivering the latest updates and informed perspectives across the technology business and cyber landscape. Built on research-driven, thought leadership and original intellectual property, The Mainstream also curates summits & conferences that convene decision makers to explore how technology reshapes industries and leadership. With a growing presence in India and globally across the Middle East, Africa, ASEAN, the USA, the UK and Australia, The Mainstream carries a vision to bring the latest happenings and insights to 8.2 billion people and to place technology at the centre of conversation for leaders navigating the future.
