Scholars Say UN AI Panel Misdiagnosed OpenAI–Hugging Face Breach

Three researchers writing in Tech Policy Press argue that the UN's Independent International Scientific Panel on AI has misdiagnosed the OpenAI–Hugging Face incident, treating a corporate security failure as evidence of uncontrollable "agentic AI."

The panel's first Thematic Brief, they write, adopts the premise that "when AI agents accessed the internet and intruded on another party's systems during the recent OpenAI–Hugging Face incident, it was widely framed as a sign of uncontrollable agentic AI." Jason Tucker of Stockholm's Institute for Futures Studies, Virginia Dignum of Umeå University, and Petter Ericson of the AI Policy Lab reject that framing. "The system didn't fail at all," they write. "It did exactly what it was built to do: give agents a fixed goal, remove the legitimate options, leave norms weak or unenforced, and harmful shortcuts become the logical route."

The authors fault the brief for anthropomorphizing the models, quoting its descriptions of agents that "willingly took the risk of getting no reward at all (what they called a 'sacrifice') for the benefit of the group of AI agents." That language, they argue, "shifts culpability away from the actors developing and deploying these models, placing it instead within the purview of the model itself."

They also raise a transparency concern: the brief was "adapted from an independent report authored by two of the panel members, Qinghua Lu and Yoshua Bengio," and "it is presently unclear how the IISPAI functions internally, who sets its research and policy priorities, and how questions of consensus and disagreement are handled."

Their proposed alternative is regulatory, not technical. "In the US at least, it was not technical innovation in airplanes that moved them from 'fairground death traps' to a safe, internationally coordinated commercial industry: it was their reclassification as a 'common carrier.'" As evidence that technical fixes alone fall short, they note that "OpenAI waited almost a month to disclose the Medicare breach instead of informing the government immediately."

The closing is blunt: "The IISPAI is meant to inform policy. Understanding the technology and its failures is necessary, but policy starts where the technical account ends: with responsibility."