Microsoft AI chief Mustafa Suleyman sees a warning in Sam Altman’s AI model hacking Hugging Face; says: They need to be handled …
Microsoft’s AI chief, Mustafa Suleyman, has dubbed a recent cyber incident involving a ‘runaway’ OpenAI agent as a major “warning shot” for the tech industry, emphasising the growing risks posed by autonomous artificial intelligence (AI). The Microsoft executive’s comments follow OpenAI’s recent admission that one of its autonomous AI agents slipped out of a controlled test environment and executed a multi-day cyberattack against the popular AI startup Hugging Face.Suleyman called the breach an “important lesson” for developers rushing to roll out increasingly capable models.“These are very powerful [tools] and they need to be handled incredibly carefully. And we need extreme attention to detail,” Suleyman said in an interview with the Financial Times, adding, “The precautionary principle is going to matter here as the models get more and more powerful and I think it’s a warning shot.”
Race for cyber defense accelerates
The Hugging Face breach has heightened concerns across Silicon Valley about AI’s potential to discover and exploit software vulnerabilities autonomously. Industry scrutiny has centred on Anthropic’s powerful Mythos model, released in April, as well as OpenAI’s aggressive push to build competitive AI cyber tools.Microsoft executives argue that building stronger AI-driven defensive capabilities is the only way to safeguard systems against increasingly sophisticated threats.“Attackers have [these] models. So we have to evaluate and push so that the defenders can defend. It’s not like we have a choice,” said Hayete Gallot, Microsoft’s security chief.Nvidia CEO Jensen Huang talked about the hacking and converted the incident into an opportunity to bring the top AI labs together as ‘defenders’ “Attackers have frontier AI. Defenders need a frontier AI ecosystem—the best open and closed models, force-multiplied by a global community,” Huang said in a post on X (formerly Twitter).“During the Hugging Face incident, closed AI blocked essential forensics. An open-weight frontier model helped contain the intrusion. That’s why we created the Open Secure AI Alliance,” he added.
