Sam Altman says AI is finding ‘Insane’ hardware-level bugs we never expected

OpenAI CEO Sam Altman has raised concerns over growing capabilities of artificial intelligence in finding security vulnerabilities beyond our expectations.

In a recent interview, Altman explained that it is becoming hard to restrict or confine AI to the sandboxes as they just work for limited time. Once the model starts evolving its cutting-edge abilities, you can’t even find the traces of how and when they escaped.

While ongoing monitoring buys time, Altman notes that true safety requires AI alignment: getting the model to internalize human values. With the risks of misalignment at an all-time high, he stresses that alignment research is still in its early stages and far from resolved.

Speaking about how traditional cybersecurity approaches may feel outdated or ineffective in front of AI’s finding hardware-level bugs.

“I don't even know a dramatic enough word to use… models can find bugs that are outside of the normal threat model of what we think a hacker can do… hardware-level, insane stuff… I no longer think we're gonna be able to just patch all the bugs.”

The CEO of ChatGPT-maker company argues that traditional and piecemeal vulnerability patching is no longer sufficient and the internet’s security infrastructure must be rebuilt using AI-powered defences.

Altman views locally run open-source AI models as a security risk, noting that even global agent monitoring cannot prevent all harmful actions.

Critics argue Altman's framing of AI as "too powerful to contain" serves OpenAI's commercial interests by shifting responsibility away from the developer when issues arise.

Altman’s recent remarks come after the PHP security vulnerabilities that hit millions of websites across the world, leaving them offline. In the midst of the crisis, millions of attempts have been performed to exploit these loopholes. And these vulnerabilities were discovered by AI.