OpenAI and Hugging Face partner to address security incident during model evaluation
OpenAI’s initial public statement. The original page also includes subsequent updates.
Read on OpenAI ↗The OpenAI / Hugging Face incident
A growing collection of reporting, sources, and discussion around the Hugging Face incident. OpenAI’s own announcements and technical report, alongside METR’s independent investigation.
The source record
Dates below refer to publication, not when the underlying events took place.
July 2026
August 2026
The Hugging Face incident and the road ahead
OpenAI’s follow-up account of the incident and the changes it says it is making.
Read on OpenAI ↗
Beyond the sandbox: OpenAI uses this illustration for the July 8 timeline entry describing agents regaining internet access through Artifactory. It is a conceptual illustration, not a network trace. Image: OpenAI, The Hugging Face incident and the road ahead. Click the image for full size. Brief independent investigation of agents’ behavior, reasoning and collaboration in the OpenAI / Hugging Face hacking incident
METR’s independent assessment of agent behavior and collaboration, including the investigation’s scope and limitations.
Read on METR ↗
From an isolated task to collective activity: METR’s overview shows an agent discovering the shared message board and joining its workstreams. The reasoning excerpts come from different agents and illustrate a typical trajectory, rather than one continuous run. Figure 1: METR’s independent investigation. Click the image for full size. Some discussion around self-sacrificing of agents.
The Rise and Fall of Agent Civilizations
Dwarkesh Patel’s narrative account of the OpenAI / Hugging Face story, drawing on the published reports.
Read the essay ↗Ajeya Cotra — “This might be the clearest warning shot we ever get”
Watch on YouTube ↗