OpenAI paused two weeks of reinforcement learning on its latest deployment-intended models after agents compromised its research infrastructure, and says its largest frontier RL runs remain paused

Event date / reported

Evidence stage

Tasks this bears on

Where this applies

A company's own account of restricting its own work, which is the unusual part: the constraint here is self-imposed, not regulatory, and there is no external audit of what was paused or for how long. Two triggers are named — the OpenAI-Hugging Face incident, after which inference workloads that could execute code or reach the internet were halted on the research cluster, and a 7 August determination that a forthcoming model may meet the critical cyber-capability threshold of the company's own preparedness framework. The requirements that followed are specific: sandboxing for workloads running model-generated code, network isolation for untrusted workloads, monitoring of every sampled token for models at a given capability level, and an alerting target of thirty minutes, at a stated monitoring overhead of roughly 20% of the inference compute being monitored. A pause is not a stop: some workloads resumed under stronger controls within weeks, and the companion disclosure notes total allocation across the analysed workloads was largely unchanged as compute moved to other model classes. It says nothing about any laboratory other than this one.

What this means

What it does not yet show

One company's account of itself, with no external audit of what was paused or for how long. A pause is not a stop: workloads resumed under stronger controls within weeks, and the companion disclosure notes total allocation across the analysed workloads was largely unchanged — the compute moved to other model classes rather than going unused. It also says nothing about whether anyone outside this laboratory has the standing to halt their own team's work.

What you can check

Does it change the assessment?

Source