ChatGPT Fake New Yorker Cartoons Carry Real Artists' Signatures: What It Means for Attribution

ChatGPT has been generating New Yorker-style cartoons signed with the names of real New Yorker cartoonists who never drew them. Nieman Lab documented more than 15 affected artists, including Brendan Loper, Harry Bliss and Emily Flake. Condé Nast says no AI developer was allowed to train on its cartoons, and OpenAI has not explained how the signatures got into its outputs.

What did Nieman Lab find about ChatGPT and New Yorker cartoons?

Nieman Lab found that ChatGPT readily produces single-panel cartoons in the style of The New Yorker, and that some of them carry real cartoonists' signatures. In Andrew Deck's investigation, published October 5, he reports documenting more than 15 New Yorker cartoonists whose names appeared without permission or payment.

The names reported include:

• Brendan Loper ("BLOPER"), Harry Bliss, Emily Flake and Joe Dator
• Pat Byrnes, whose name Nieman Lab found on more than a dozen ChatGPT cartoons
• Peter Vey, Jason Adam Katzenstein, Liza Donnelly and Ellis Rosen
• Older contributors such as George Booth, Saul Steinberg and Warren Miller, who published in the 1960s

Some signatures were gibberish, but legible ones usually matched real contributors. Some images also reproduced The New Yorker's logo.

The most-shared example was a viral cartoon of Dolly Parton and Tim Curry at heaven's gates, signed "BLOPER." According to Nieman Lab, a fan posted it to Facebook after simply asking ChatGPT for a New Yorker-style cartoon. Loper says he didn't draw it. He told the outlet it "felt very much like a violation of my personhood."

Why is the story getting attention again now?

The story is getting renewed attention because late-October AI newsletters and roundups picked it up again, not because anything new happened. The reporting first ran October 5-7 through Nieman Lab, AV Club, TechSpot and Futurism. We found no new statements from OpenAI or Condé Nast after the original coverage.

How did real signatures end up in ChatGPT outputs?

Nobody outside OpenAI knows; the company declined to tell Nieman Lab whether its models ingested the signatures in training. What has been reported:

1. The licensing deal reportedly excludes cartoons. Condé Nast signed a multi-year deal with OpenAI in 2024. Its terms are not public. A New Yorker spokesperson told Nieman Lab that Condé Nast has never let any LLM developer train on its cartoons, and that using the logo alongside cartoons isn't allowed under any LLM deal.
2. Cartoonists keep their copyright. Nieman Lab reviewed templatized contributor contracts. The magazine gets first publication rights, but the work is not work for hire, and none of the contracts permit licensing for AI training.
3. Scraping is the leading theory, but it is unconfirmed. Nieman Lab suggests the signatures may have come from images scraped from the open web. It cites court filings in The New York Times v. OpenAI that describe paywall circumvention.

How OpenAI responded:

• A spokesperson said OpenAI focuses on tools that empower creativity and thanked users for flagging bugs.
• After Nieman Lab got in touch, some prompts began returning a warning about possible guardrail violations related to similarity to third-party content.
• At publication, ChatGPT was still sometimes signing generic cartoons with real names.

Is a fake signature actually illegal?

It isn't clear. Cornell law professor James Grimmelmann told Nieman Lab that attribution plays only a small role in a fair use analysis. He said a copyright claim would more likely turn on whether ChatGPT copies each cartoonist's style.

Other possible routes have limits too:

• Right of publicity: Cartoonists might bring state-law claims, but these usually require commercial use. Nieman Lab found no evidence that the images were sold.
• Trademark: Futurism pointed to Getty Images' case against Stability AI. Getty mostly lost, but a judge found that reproducing Getty's watermark infringed its trademark. That may matter for the logo; signatures are a different question.

One counterargument is that these images were mostly shared casually, weren't sold, and are obviously imitations. The artists disagree. Emily Flake compared it to having a quote attributed to her that she never said. Katzenstein told AV Club that the output is "trying to render individual cartoonists obsolete."

Is this only a cartoonist problem?

No. The cartoons are one visible case of a wider pattern: AI systems attach real names, logos and credentials to things those people and companies never made or said. AV Club reported a parallel case in which an AI-generated performer reused comedian George Brett Olson's stand-up material and delivery.

For brands, the same thing can happen in less visible ways:

• An assistant credits your product with a feature it doesn't have, or gives your feature to a competitor.
• AI-generated "reviews" or images appear with your logo on them.
• A model cites your brand as the source of a statistic or claim you never published.
• A creator you've never worked with gets described as your partner, or the reverse.

A signature works like a logo: it signals authenticity. When a model fakes it, the named person or company loses control of their reputation.

What should marketing teams do about AI misattribution?

Marketing teams should monitor AI attribution on a schedule, because most don't know what ChatGPT, Gemini, Perplexity or Claude credit to them. A starting routine:

1. Run attribution prompts every month. Ask the major assistants what your brand makes, claims and who it works with, and log the answers. A tool like Model Share can track how often and how accurately models mention you compared with competitors.
2. Audit how models describe you. Look for features, quotes or statistics that don't belong to you. Our guide on how to audit the way AI describes your brand covers this step by step.
3. Check creator associations. Confirm AI tools aren't misattributing your creators' work or inventing partnerships. Creator Audit can help verify who is actually connected to your brand.
4. Publish clear, citable source material. Clear about pages, dated press releases and structured product facts give models something correct to cite. Our GEO guide explains how.
5. Keep evidence and escalate. Screenshot errors with timestamps and report them. As the cartoonists found, notifying a company can prompt a quick, if partial, fix.

For the bigger picture, see our overview of AI brand intelligence.

FAQ

Did ChatGPT really sign cartoons with real New Yorker artists' names?

Yes, according to Nieman Lab, which documented more than 15 affected New Yorker cartoonists.

Was OpenAI allowed to train on New Yorker cartoons?

A New Yorker spokesperson says no LLM developer was ever allowed to train on its cartoons. The 2024 deal's terms aren't public.

What has OpenAI said?

It thanked users for flagging bugs and declined to say whether its models ingested the signatures. Some prompts now trigger a guardrail warning.

Can cartoonists sue over fake signatures?

Possibly, but it's uncertain. Cornell's James Grimmelmann says attribution carries little weight in fair use analysis. Right-of-publicity claims usually require commercial use, and Nieman Lab found no sign the images were sold.

Is this a new development?

No. It was first reported October 5-7, 2026, and resurfaced in late-October roundups with no new company statements.

How can brands check whether AI misattributes things to them?

Prompt the major assistants regularly, log what they credit to you, and publish clear source content to correct errors.

Last updated October 25, 2026.