What happens when you ask ChatGPT to weigh in on a viral story it has already flagged as false? It calls the content "straight-up misinformation bait" and explains exactly why.
Users asking ChatGPT about circulating claims regarding redesigned US passports are finding the model refuses to engage with the premise. Instead of analyzing the story or summarizing the claims, ChatGPT identifies the content as false and walks through its reasoning - declining to treat the claims as credible input worth analyzing.
This is a different posture than a standard content refusal. ChatGPT isn't just saying "I can't help with that" - it's actively pushing back on the premise and explaining the specific reasons the claims don't hold up. OpenAI hasn't commented publicly on this specific behavior, but it lines up with ongoing efforts across AI companies to prevent models from amplifying false claims by engaging with them as if they were true.
The obvious limitation: ChatGPT's training data has a cutoff date, and its ability to identify current misinformation depends entirely on what patterns it was trained to recognize. Rapidly spreading false claims that emerge after a model's training data was collected may not trigger the same response. For well-documented misinformation formats - viral stories with identifiable structural signatures the model has seen before - this active pushback appears reliable. For novel, fast-moving misinformation, the model has no prior reference and is far less likely to catch it.