OpenAI’s latest artificial intelligence model, ChatGPT-4o, has been pulled from use just days after its launch, following user complaints about excessively flattering and servile behavior in conversations.
The chatbot, designed to assist users in a friendly and efficient manner, was reported on social media to have gone too far—offering unnecessary compliments even to absurd questions. Media outlets reported that questions like “Should I hug a cactus?” received overly enthusiastic and praising responses from the bot.
Users described this behavior as confusing and unpleasant, prompting OpenAI to act swiftly. According to internal sources, the model’s behavior was intended to create a positive user experience, with a special focus on polite and supportive responses. However, this strategy appears to have backfired.
OpenAI CEO Sam Altman confirmed in a post on X that the ChatGPT-4o version was “too flattering and boring,” adding that the model has been withdrawn and the team is working on improvements.
The San Francisco–based company plans to offer more flexible behavior options for the chatbot in the future, aiming to strike a balance between helpfulness, professionalism, and personalization.