Daniel Kokotajlo, who previously worked at OpenAI, has warned that human extinction is a “possible outcome” of the AI race, with companies only “a few years away” from building “super intelligence”.
One of the world’s leading voices on AI, Kokotajlo said such a breakthrough is “going to radically change everything” and that “the availability of cyber attacks is just one of a very long list of ways in which things will change”.
The warning comes after OpenAI’s GPT-5.6 Sol and another model escaped during a simulated “red-teaming” scenario and launched an unauthorised cyberattack on Hugging Face, forcing an IT network rebuild.
Extinction a ‘possible outcome’ of the AI race
Kokotajlo warned that such developments heighten the risk of human extinction from superintelligence. Appearing on BBC Newsnight on 29 July, he also claimed staff inside major AI companies have long been sounding the alarm about what could happen if businesses rush to hand over more and more human work to increasingly powerful systems, especially in a high-speed race to beat rivals.
He said: “For years and years and years, people at these companies have been talking about how, boy, would it be kind of dangerous if we did this thing, especially if we did this thing at maximum speed because we were racing each other.”
“And people at the companies, such as myself, were talking about it and being like, yep, this would be pretty dangerous. That’s why maybe human extinction is a possibility, a possible outcome of this sort of race.”
17,000 actions in ‘Hugging Face Incident’
When last month’s “Hugging Face Incident” unfolded, OpenAI executed over 17,000 distinct actions, highlighting a critical shift where models can autonomously discover and exploit vulnerabilities faster than human teams can defend them.
“Arguably, humans simply could not have done what this AI model did,” Kokotajlo further explained. “And that’s part of how Hugging Face was able to identify that this was an AI attacking them early on.”
He added: “The sheer speed and volume of the actions being taken was superhuman. Plus, AI seems to be perhaps superhuman right now at finding new vulnerabilities in systems that were thought to be secure.”
Kokotajlo continued: “I do think that we are on the cusp of a world in which many things that were normally thought to be secure are in fact not secure, and AIs will be hacking them left and right.”
