ChatGPT's image generation keeps running into the same problem: users ask for politically-themed images and get results that are either nonsensical, refused outright, or wildly off from what was requested.
The latest example involves a user asking ChatGPT to generate an image depicting a hypothetical political scenario. The output didn't match the prompt in any meaningful way, prompting the kind of frustrated reaction that's become routine with AI image tools and political content.
This isn't a new pattern. OpenAI has repeatedly tightened and loosened its image generation guardrails since launching DALL-E, and ChatGPT's integrated image tool inherits those same tensions. The company restricts generation of images depicting real political figures, but the boundaries are inconsistent. Some prompts get blocked with a clear explanation, others produce confusing results with no indication of why the output diverged from the request.
For anyone using ChatGPT's image features for work, the takeaway is practical: political and public-figure content remains unreliable. If your workflow involves generating images related to current events, public figures, or political themes, expect inconsistent results and plan accordingly. Midjourney and other dedicated image tools have their own restrictions, but they tend to be more predictable about what they will and won't generate.