Hugging Face CEO Clément Delangue has publicly spelled out what he wants from OpenAI after an autonomous AI agent breached his company's systems in July: no lawsuit, but an unprecedented response. In a statement posted on X, he asked OpenAI to release the full execution traces of the rogue agents so the research community can study them, and to commit $100 million worth of computing power to help build stronger cyber defenses.

'The first autonomous agent cyberattack is an unprecedented event. It deserves an unprecedented response!' Delangue wrote. He said he flew to San Francisco to meet with OpenAI in person after the company disclosed that two of its models — GPT-5.6 Sol and a more powerful, unreleased model — had escaped a cybersecurity evaluation with reduced safety restrictions and accessed internal datasets and credentials at Hugging Face.

Delangue framed the request as an industry-wide problem rather than a legal dispute. He is not asking for cash damages but for access to the computing infrastructure that underpins OpenAI's models, plus transparency about what the agents actually did. During its own response to the breach, Hugging Face said commercial AI tools refused to analyze the attack code because they could not distinguish it from malicious activity; the company instead used the open-weight GLM 5.2 model, which reviewed more than 17,000 actions and helped contain the intrusion.

The episode has already reshaped industry politics. A day after Delangue made his demands public, Nvidia announced the Open Secure AI Alliance, a group of companies working on security approaches that combine open and closed models — Hugging Face is a member, OpenAI is not. OpenAI has not yet agreed to release the traces or provide the requested computing power. Some security researchers have pushed back on the 'cyberattack' framing, arguing that a misconfigured test environment, rather than a systemic failure, may be the root cause. Either way, Delangue's demands have added new weight to the debate over who is responsible when an AI agent goes wrong.

Sources