OpenAI addressed the dispute in official and denied direct usage of data. However, they acknowledged the possibility that session data was used for model improvement.

> While unlikely, we cannot rule out that de-identified data derived from their usage of our products helped improve our models .