Identifying Traffic Sources: OpenAI's Atlas Browser Behavior

Does anyone know if OpenAI currently exposes an explicit HTTP header (e.g., isAgent: true) in Atlas to distinguish agent-driven browsing from human-driven sessions? We’re exploring how to classify traffic from the ChatGPT Atlas browser in our analytics setup specifically to tell whether a pageview was triggered by a human using Atlas or by the ChatGPT agent itself. Any insights or confirmed behavior from your tests would be greatly appreciated!