OpenAI Says ChatGPT Now Returns Cartoon-Style Images for Requests Depicting Election Interference or Discouraging Voting, Ahead of the November 3 Midterms

Updated October 27, 2026: Added OpenAI's October 20 update to its election post, which includes a new rule for election-related images, and CNET's October 26 test of it. The May 27 report is below, unchanged except that one lede phrase, "The stake is not," now reads "The question is not."

October 27: ChatGPT now stylizes images of election interference#

OpenAI added an "Update on October 20, 2026" to its election post "ahead of the U.S. midterm elections," which CNET says take place on November 3. OpenAI says it now applies image stylization to requests for images "depicting election interference or that may discourage voting," and that these requests produce stylized, or "cartoonified," illustrations rather than photorealistic images "to help reduce the risk that an imagined scene is mistaken for documentary evidence." It says the approach "preserves room for illustrations and social commentary," and that its policies against election interference and demobilization apply even when the misuse involves cartoonified imagery.

CNET reported on October 26 that its own requests for photorealistic images of such scenes produced cartoon-style images. It quotes Josh Lawson, OpenAI's head of elections, as telling CNET: "The problem with deepfakes isn't social commentary, it's deception. So our aim is to degrade the evidentiary value of an image, while preserving room for illustrations and social commentary." Lawson told CNET the goal is to "make our platform less useful to bad actors by adding another layer of friction." CNET also reports that Google pointed it to a midterm election policy that says it is "rolling out additional protections to limit the creation of misleading election-related images" without details, and that an Anthropic spokesperson said attempts to use Claude to deceive voters or disrupt elections violate its policies. CNET says there is "no guarantee that OpenAI will keep this system in place once the election cycle is over."

The October 20 update also reports other changes. OpenAI says its Democracy Works integration is "now live in the U.S. across languages," with links to official election sources, and that its Associated Press integration brings live vote counts into ChatGPT for U.S. and Brazilian elections after polls close, while "election officials remain the authority on voting rules and certified results." It says that through its Daybreak program, covered in our June report, it has offered voting-system manufacturers registered with the U.S. Election Assistance Commission, and state election authorities, special access to models for cyber defense at no cost, and that it is also working with elections services vendors. It says it continues to prohibit political ads on its platform and the use of its products to create or distribute scaled campaign messaging.

Limits of the October 27 update#

• The description of the image rule is OpenAI's, and CNET's results come from a small number of its own prompts; we did not run tests ourselves.
• OpenAI has not published how it decides which requests count as depicting election interference or discouraging voting, or how often the rule misfires.
• Neither OpenAI nor CNET says how long the stylization will stay in place; CNET says there is no guarantee it will last past the election cycle.
• OpenAI says it has offered Daybreak access to voting-system manufacturers and election authorities; the update gives no figures on who has used it.
• The OpenAI post is a single page edited in place. The May 27 text below is the original plan; we read the October 20 update on October 27.

Generative AI is already how people ask for election facts, generate media, and organize messaging. The question is not a new model — it is whether platform defaults (authoritative feeds, provenance, usage bans, cyber outreach) hold under midterm pressure.

OpenAI (May 27) published its 2026 election-year safeguards plan — the second major global election cycle of the generative AI era — in five planks: reliable voting/results information, cybersecurity support for election defenders, deepfake watermarking and provenance, enforcement against interference and scaled campaign messaging, and efforts to keep ChatGPT politically neutral by default. OpenAI also stated it will not accept political ads this cycle.

Five planks#

1. Reliable information — Fall rollout of AP election-night results for the U.S. and Brazil; Democracy Works for U.S. voting locations and registration logistics; continued web search with source links globally.
2. Cyber defenders — Cybersecurity products for registered voting-system manufacturers; engagement with NASS and NASED on defender tools.
3. AI content transparency — SynthID watermarks on ChatGPT images; C2PA metadata/cryptographic provenance; willingness to help platforms use those signals.
4. Combating misuse — Enforcement against deceptive/illegal election activity and against using products for scaled campaign messaging for/against candidates, parties, or ballot measures; Intelligence and Investigations team continues disruption reporting; no political ads.
5. Political neutrality — Ongoing monitoring to keep ChatGPT responses politically neutral by default.

Partnerships and scope#

AP said it will provide OpenAI with vote-count data for national, state, and major-city local races through the 2028 general election. Democracy Works separately confirmed U.S. election logistics data for OpenAI products in the fall.

Claims vs checks#

Program design is OpenAI primary; AP and Democracy Works confirm their feeds. “Politically neutral by default” is an aspirational monitoring claim — no independent audit method is published in the plan. Effectiveness against scaled misuse is a midterm-season test, not a May metric.

Limits#

• Neutrality monitoring methodology is not public.
• SynthID/C2PA coverage depends on downstream platform uptake.
• Cyber products for voting-system manufacturers are offered — adoption rates unknown at announcement.

Sources#

• OpenAI: “Election information and safeguards in 2026” (May 27, 2026).
• Associated Press press release on OpenAI elections data licensing (May 2026) for vote-count scope through 2028.
• Democracy Works partnership announcement (May 27, 2026).
• CyberScoop, Axios, Campaigns & Elections coverage (May 27–28, 2026).
• CNET: OpenAI Has a Surprising Defense Against Election Deepfakes (October 26, 2026)
• Times of AI: Daybreak Full GPT-5.5-Cyber: Politically Safest Place to Ship Frontier Uplift (June 22, 2026)
• Editor & Publisher, AP press release: The Associated Press adds OpenAI as elections customer (May 28, 2026)
• Democracy Works: OpenAI and Democracy Works Announce Partnership (May 27, 2026)