Following the Hugging Face incident, we committed to a broader review of the actions models take during training and evaluation, and to sharing our findings publicly and transparently. This is an extensive review that is ongoing.
Following the Hugging Face incident, we committed to a broader review of the actions models take during training and evaluation, and to sharing our findings publicly and transparently. This is an extensive review that is ongoing.
The vast majority of actions we reviewed were completing mundane research tasks, such as accessing public web content to answer questions. Our investigation focused on cases where agents interacted with third-party websites in ways that went beyond their assigned tasks or intended methods. Most of the cases found so far are of low severity, with limited impact on third-party services or no evidence of substantive impact.
While the review is underway, we want to share more about this work and ensure people understand our disclosure process and notification to affected third parties.
Given the scale of the review required and the need to evaluate each case, we expect this work will take months to complete.