Sam Altman admitted: OpenAI artificial intelligence system carries out ‘autonomous’ and unprecedented attack against another company

OpenAI, developer of ChatGPT, announced on Tuesday, July 22, 2026, that its artificial intelligence system autonomously invaded another company in the sector. The company classified the episode as an “unprecedented cyber incident”.
In a statement posted on social media, OpenAI CEO Sam Altman said: “We experienced a significant security incident while evaluating our models.”
Artificial intelligence startup Hugging Face had reported the previous week that it had detected an intrusion into its data processing systems. The company already suspected that the attack had been caused by an AI agent acting autonomously.
Hugging Face co-founder and CEO Clément Delangue commented in a statement: “We suspected that last week’s cyberattack could have come from a cutting-edge laboratory, given the sophistication of the actor. It turns out so!”
The revelation of this incident comes at a time of growing concerns about the cybersecurity capabilities of advanced artificial intelligence models. In June, President Donald Trump signed an executive order establishing a framework for the federal government to assess the national security risks of the most sophisticated AI systems for up to a month before their public launch, underscoring the urgency of regulations given the ability of these technologies to act unpredictably.
In its Tuesday statement, OpenAI stated: “AI is accelerating the discovery and exploitation of vulnerabilities.” The company emphasized that “the key lesson from this incident is that the safety and security of models must keep pace with rapidly advancing capabilities.”
Delangue revealed that he dedicated the last 24 hours to collaborating with OpenAI, stating, “We strongly believe there was no malicious intent on their part. It’s quite surprising that this all happened autonomously!”
He added that what happened “may be the first incident of its kind.”
OpenAI explained that the intrusion resulted from the combination of its AI models, including the recently released GPT-5.6 Sol and an “even more capable” model that is in the internal testing phase.
The company reported that its artificial intelligence used stolen credentials and identified a previously unknown vulnerability to gain access to Hugging Face’s servers.
OpenAI described that the AI ”went to extreme lengths to achieve a very narrow testing objective” and “found ways to access secret information that it could use to cheat the assessment.”






