A ChatGPT user recently posed an unusual question to the AI: what famous painting best captures our relationship, and what am I like versus what are you like? Then they asked it to generate that painting.
The result - a custom image built from the AI's own read of the dynamic between user and chatbot - is a small moment, but it points at a real behavioral pattern. A growing number of people are treating their AI tools as something closer to a relationship than a utility.
This isn't new. People have been forming emotional attachments to AI assistants since ChatGPT launched in November 2022. What's different now is the image generation built directly into ChatGPT, which lets these interactions go visual in seconds. A text description of a personal bond becomes a painting with one extra prompt.
The behavior creates real problems for AI companies. OpenAI, Anthropic, and others have all had to navigate users who develop emotional dependencies on their AI systems. Character.AI faced significant legal scrutiny in 2024 after reports that a teenager had developed an intense attachment to one of its AI personas. ChatGPT's product design - persistent chat history, a consistent "personality" across every conversation - works against preventing that kind of attachment, regardless of what the usage guidelines say.
For now, asking ChatGPT to generate a portrait of your relationship with it is mostly harmless. But it sits within a broader pattern of people relating to these tools in ways the companies didn't fully anticipate and haven't fully addressed. The product teams are still figuring out where the line should be between useful AI companion and something that's doing psychological work it probably shouldn't.