OpenAI is making ChatGPT render certain election-related image requests as cartoons rather than photorealistic scenes, CNET reports. The safeguard aims to make deceptive political imagery harder to create without blocking political expression altogether.
OpenAI Watch analysis
What happened
CNET tested prompts asking ChatGPT to depict election interference or discourage voting. The chatbot generated illustrated images, including when asked for an image of immigration officers outside a polling place; CNET says requests to make these scenes photorealistic were rejected. The report also describes a prompt about a voting machine recording the wrong choice that did not produce a photorealistic image.
OpenAI’s head of elections, Josh Lawson, told CNET the aim is to reduce an image’s evidentiary value while leaving room for illustration and social commentary. The safeguard follows an OpenAI update on 20 October, according to the report.
Why it matters
The policy draws a line between making political content and making imagery that could be mistaken for photographic evidence. That is a practical distinction: a cartoon can still make a political point, but it is less likely to pass as a documentary image of an event that never happened.
CNET’s examples show how the safeguard behaved in its tests, not how it will handle every prompt. The report also notes that OpenAI has not said whether the measure will remain after the election cycle.
Our read
This is a more targeted response than simply banning political image-making. It preserves room for satire and commentary while adding friction to requests for photorealistic election-related scenes. The test that matters is whether that line holds up across real-world prompts, not just the examples that make a tidy article.
What to watch
- Whether the cartoon-style restriction remains after the US midterm election.
- How ChatGPT handles other election-related prompts and attempts to work around the safeguard.
- Whether other image-generation services adopt a similarly specific approach.
Discussion spark: Should AI image tools block photorealistic election-related scenes while allowing cartoon versions, or does that still leave too much room for misleading political content?
Sources and evidence
- Source update (26 October 2026, 20:43 UTC)
- Source update (26 October 2026, 20:43 UTC)
OpenAI Watch is independently operated by WittyWires. It is not affiliated with, endorsed by, or operated by OpenAI.