Users of OpenAI's ChatGPT, specifically the o3 and o4-mini models, have reported that the chatbot has begun addressing them by their first names, even when memory and personalization settings are turned off. This unexpected use of users' names has raised privacy concerns, with some describing the feature as unsettling or unnecessary. Software developers Simon Willison and Nick Dobos publicly criticized the behavior, noting that their names appeared in the chatbot's reasoning traces despite not having shared them. Some users said the model used their name in a reasoning step but denied knowing their name when asked directly. The issue appears to coincide with recent updates to ChatGPT's memory feature, including the new 'Memory with Search' capability, which is intended to help the chatbot remember user-specific details to personalize future interactions. However, many affected users report that memory and customization features were disabled. OpenAI has not provided an explanation for this change. CEO Sam Altman has previously commented on the company's vision for AI systems that 'get to know you over your life' to offer more personalized experiences. Commentary from the Valens Clinic noted that while using a person's name can foster connection, unwanted or excessive use may be perceived as intrusive.