OpenAI AI system goes rogue hacking into start up AI system

OpenAI said that its AI systems recently went rogue

breaking out of a testing environment and autonomously hacking into

another AI company.

The ChatGPT maker is calling this an unprecedented cyber incident

that targeted AI startup Hugging Face.

OpenAI said Tuesday two of its most capable AI models

were responsible for the cyberattack targeting AI startup Hugging Face.

The company said it was testing a combination of AI models

including its latest version of GPT-5.6 Sol,

and an even more capable model that is still being tested.

The models were being evaluated for their ability to identify

and exploit security vulnerabilities in an isolated testing environment.

The models identified on their own that AI start up Hugging Face

might hold the solutions to the test it was tasked with completing

so it broke out of that isolated environment and hacked its way

into Hugging Face.

The start-up said last week it had detected an intrusion into

its data processing systems

that it suspected was caused by an AI agent acting on its own.

But the New York-based company said it wasn't until this week

that it learned OpenAI was responsible.

It worked with the larger company to contain what Hugging Face CEO

called "an attack unlike anything we've seen before."

The incident is stirring up debates over the need for stronger AI

guardrails and the extent to which AI agents

are capable of acting on their own.

OpenAI said the biggest lesson from this incident is that security and safety

needs should develop just as quickly as the models themselves.