ARTICLE · LEARNING · 3 MIN READ
Oxford Dean Soumitra Dutta on How Claude Mythos Is Changing AI Conversations
Soumitra Dutta, former dean of Oxford Said Business school and AI scholar, says, "It (Claude Mythos) signals a shift from AI as a tool

Claude Mythos, Anthropic's latest AI system and most powerful model, has been making news. "During our testing, we found that Mythos Preview is capable of identifying and then exploiting zero-day vulnerabilities in every major operating system and every major web browser when directed by a user to do so," the firm wrote in a recent post on its website. In his recent post on X, Soumitra Dutta, former dean of Oxford Said Business school and AI scholar, says, "It (Claude Mythos) signals a shift from AI as a tool to AI as an autonomous actor." This is not a smarter chatbot. It will enable cyberattacks to be carried out by individuals who are not themselves technical experts.
Claude Mythos represents an entirely different class of capability. "This is about operational intelligence at scale," says Dutta, who's co-creator of the Network Readiness Index and the Global Innovation Index.
According to mindstudio.ai, Claude Mythos does "what a top-tier human security researcher does — but Claude Mythos can do it faster, at scale, and without getting tired". The model has been restricted to a limited group of industry partners and open source developers. Claude Mythos triggered Anthropic's internal safety standards under the firm's Responsible Scaling Policy. There are a few realities leaders can't ignore, argues Dutta, who has a PhD in computer science with a major in AI from the University of California, Berkeley. "Governance is lagging," he says. "Mythos is being tightly restricted—not because it failed, but because it worked too well. What happens when similar capabilities become widespread?"
Dutta -- who is author/coauthor of books such as Throwing Sheep in the Boardroom -- makes you think. Most leadership conversations are still framed in terms of adopting and deploying AI‚ rather than controlling it․ How do we scale it? How do we stay competitive? These are good questions‚ but they are increasingly the wrong questions․ What leaders need to be asking: What can we let AI do‚ and how do we govern that boundary? In that respect‚ AI would push organisations away from procurement thinking and more towards constitutional thinking. What authority does an autonomous system have? Who is liable when it acts? And how is that accountability enforced across a supply chain‚ a regulatory environment‚ a geopolitical landscape? "AI just crossed a line, and most leaders are not ready," says Soumitra Dutta former oxford dean, who is cofounder of the Portulans Institute, a research think tank in Washington DC.
Loading the full page…