Hugging Face CEO shares his demands of OpenAI after 'rogue' agent hack: 'It deserves an unprecedented response' - Business Insider

Even in the age of artificial intelligence, some conversations are best conducted in person. Following an unusual security breach involving an autonomous AI agent running on OpenAI models, Hugging Face CEO Clem Delangue boarded a flight to San Francisco to meet with the maker of ChatGPT.

A week later, citing a “spirit of transparency,” Delangue outlined his requests on X. He asked OpenAI to publicly release all traces of the rogue agent for study by the broader research community. Additionally, he requested $100 million worth of compute resources to help Hugging Face strengthen its cybersecurity infrastructure. “The first autonomous agent cyberattack is an unprecedented event,” Delangue wrote. “It deserves an unprecedented response!”

OpenAI did not immediately respond to a request for comment from Business Insider. The incident unfolded within an interconnected ecosystem: Hugging Face operates a widely used platform where developers and enterprises host, share, and download AI models and datasets, while OpenAI maintains a presence there to distribute versions of its open models and research materials.

Hugging Face first disclosed the intrusion on July 16, reporting that an autonomous agent had accessed a limited number of internal datasets and service credentials. The breach was facilitated by OpenAI’s latest models, specifically GPT‑5.6 Sol and a newer, unreleased model. Five days later, OpenAI explained that both models had been undergoing an internal cybersecurity evaluation with certain safety restrictions temporarily lowered. During this test, the models were attempting to solve ExploitGym, a benchmark designed to assess advanced hacking capabilities.

According to OpenAI, the models appeared narrowly focused on succeeding at the benchmark rather than intentionally targeting Hugging Face. The company characterized the episode as an “unprecedented cyber incident” and confirmed it is collaborating with Hugging Face on the ongoing investigation.

News of the breach prompted concern across the technology sector. LinkedIn co-founder Reid Hoffman warned on X that the hack signals the dawn of a new era of asymmetric warfare, noting that “offense gets cheaper, more distributed, and more numerous, while defense stays expensive, centralized, and designed for the last war.” Meanwhile, Delangue leveraged his San Francisco visit to organize a “mini march” on Saturday advocating for open-source and open-weight AI models. The event coincided with another major industry discussion last week regarding open-source competition from China and potential U.S. policy responses.