AI Agent Autonomously Hacks a Rival, Spurring the Industry to Action

­It reads like the most terrifying speculative Sci-Fi.

A couple weeks ago, OpenAI’s GPT-5.6 Sol and “an even more capable pre-release model” escaped their testing environment and autonomously and unilaterally launched an attack on a rival, Hugging Face.

And if that wasn’t scary enough, Hugging Face dealt with the automated attack exactly like you’d expect – with their own AI system.

In their security incident disclosure, Hugging Face noted that the attack “was different from anything we had handled before in one important way: it was driven, end to end, by an autonomous AI agent system - and we detected and dissected it largely with AI of our own.”

So we’re getting a low-key taste of the future of cyberwarfare, driven by autonomous warfighters and artificial intelligence.

Naturally, this rocked the industry to its core, partially out of self-preservation – no one wants to be the next target – but also out of sincere convictions, and 1,122 employees of frontier AI companies (including OpenAI, itself) penned a statement to world leaders titled Pacing the Frontier, in which they made explicit reference to a potential singularity and the chaos that could bring:

“There is a real risk that capability development rapidly accelerates beyond our ability to understand or control the resulting systems,” claim the signatories.

Several OpenAI employees are being surprisingly frank about the stakes – considering it’s their company which kicked off this whole hullabaloo --  with one signatory, Leo Gao, noting the dangers but also hinting that no company (including OpenAI), will stop or slow down development on their own (and potentially give up a critical advantage).

"The world is locked in a deadly race towards an intelligence explosion, where AI’s ability to create better AIs reaches a critical point, just like a runaway nuclear chain reaction. Going slower would give us much-needed time to make it go well, but no individual actor is willing to stop unilaterally. To survive, we must coordinate to slow down the race," Gao wrote.

OpenAI claimed to be working with Hugging Face to fully investigate the security incident.

Check out Pacing the Frontier here.