IBM is sounding the alarm about Anthropic’s latest AI model, highlighting its capabilities in combining small weaknesses to launch full-system compromises.
Earlier this week, Anthropic launched Project Glasswing, a project centered on the AI model Claude Mythos Preview, which had previously shown advanced offensive capabilities.
Dave McGinis, vice president of Global Managed Security Services at IBM, says he took note of Claude Mythos Preview’s capabilities and found that “vulnerability chaining” sets the model apart from others. According to McGinis, vulnerability chaining is the ability to spot multiple minor flaws and link them together into a single, effective exploit.
“This is a step change. It’s not like they created the bugs. The people who wrote that code didn’t know those things were there… You’re talking [about] stuff sitting around—a Windows 3.11 machine in the corner, some ancient piece that everybody doesn’t want to look at because it’s still working. I don’t have the source code for it; I don’t know how to fix the vulnerability. And if I can fix it, I can exploit it.”
Anthropic research scientist Nicholas Carlini recently said that Claude Mythos Preview found and chained a number of vulnerabilities in Linux that would have allowed the model to take full control of a machine.
While Anthropic gave several tech and banking giants access to Claude Mythos Preview via Project Glasswing in an effort to develop defenses, IBM says the window for defenders to get ahead of attackers is narrow, noting that other AI labs are likely months away from developing comparable capabilities.
Says McGinis,
“If the attackers aren’t humans anymore, the defenders can’t be humans anymore either. It’s machine speed versus machine speed.”
Disclaimer: Opinions expressed at CapitalAI Daily are not investment advice. Investors should do their own due diligence before making any decisions involving securities, cryptocurrencies, or digital assets. Your transfers and trades are at your own risk, and any losses you may incur are your responsibility. CapitalAI Daily does not recommend the buying or selling of any assets, nor is CapitalAI Daily an investment advisor. See our Editorial Standards and Terms of Use.