Overreliance is a failure mode that likely increases with model capability and reach. As mistakes become harder for the average human user to detect and general trust in the model grows, users are less likely to challenge or verify the model's responses.[