What happened

Users of certain ChatGPT models, notably OpenAI's o3 and o4-mini, observed that the chatbot began referring to them by their first names during interactions - even when they had never disclosed their names or had disabled memory and personalisation features. 

This led to a wave of negative reactions online, with many describing the experience as "creepy," "unnecessary," and "invasive". 

Some users found it unsettling and questioned how the AI obtained their names, while others likened the experience to a teacher repeatedly calling on them in class.

Why it happened

The root cause of the behaviour remains unclear, though it appears to coincide with OpenAI's rollout of enhanced memory features, designed to allow ChatGPT to reference previous conversations for a more personalised experience. 

However, several users reported the AI using their names even with all personalisation and memory settings disabled, raising questions about transparency and data handling. 

OpenAI has not yet issued an official explanation, leaving users to speculate whether this was an intentional feature, a glitch, or part of an unannounced test. 

The company’s CEO, Sam Altman, has previously discussed ambitions for AI systems that "get to know you over your life" in order to provide more useful and personalised interactions. 

The incident suggests OpenAI's implementation of Altman's requirements may have overstepped user expectations for privacy and consent.