When OpenAI’s ChatGPT is asked to create a “New Yorker-style cartoon,” it doesn’t just copy the magazine’s visual style. According to reporting from Nieman Lab, the chatbot has been generating fake cartoons and then signing them with the actual pen names of real New Yorker cartoonists, falsely attributing AI-generated images to working artists who had no part in creating them. The practice has already gone viral at least once, and it raises uncomfortable questions about how image generators handle identity, authorship, and consent.

How a viral Dolly Parton cartoon exposed the problem

The issue surfaced publicly after the deaths of Dolly Parton and Tim Curry in late August. A fan used ChatGPT to generate a cartoon imagining the two entertainers arriving at heaven’s gates together, in the style of a New Yorker cartoon. The image spread widely, with one version on Twitter receiving 25,000 likes and additional copies circulating on Facebook, Instagram, and Bluesky.

In the corner of the image was “BLOPER,” the signature used by real New Yorker cartoonist Brendan Loper. Loper had never drawn or signed the cartoon. He found out about it only when friends, strangers, and even his twin brother’s friend began contacting him, asking if the viral image was his work. The woman who originally posted it confirmed in her comments that she had simply prompted ChatGPT to produce “a New Yorker-style cartoon” and the signature appeared on its own.

Why this is different from typical AI training controversies

Much of the public debate over generative AI and creative work has centered on whether models were trained on copyrighted material without permission. Loper told Nieman Lab he already assumed his cartoons were “out there, getting harvested” for AI training data. But having his actual signature reproduced and falsely attached to work he didn’t make felt like a different, more personal kind of harm: an identity issue, not just a copyright one.

A cartoonist’s signature functions as more than a name. Because cartoons are frequently stripped of context and reshared across social media without credit, signatures act as a kind of watermark, Loper explained, tying an image back to its creator and serving as a “certificate of authenticity.” When ChatGPT reproduces that signature on a fabricated image, it severs that link in the worst possible way: it keeps the mark of authenticity while attaching it to something the real artist never made or approved.

This is not an isolated incident

Loper’s case wasn’t the first time his signature appeared in AI output. In May, a stranger warned him over Instagram that his pen name had shown up in a ChatGPT-generated image. Over the summer, he found more examples on a Reddit forum dedicated to ChatGPT, where users were posting similar New Yorker-style cartoons they had generated themselves.

The problem extends well beyond one artist. Nieman Lab’s reporting identified more than 15 New Yorker cartoonists whose signatures have appeared on AI-generated images without their permission or any compensation, including prominent names like Harry Bliss, Emily Flake, Joe Dator, Pat Byrnes, Peter Vey, and Jason Adam Katzenstein. In its own testing, the publication got ChatGPT to output cartoons bearing the signatures of some of the magazine’s most historically significant contributors, including George Booth, Liza Donnelly, Ellis Rosen, and the late Saul Steinberg, one of the most influential cartoonists in The New Yorker’s history.

What it means for artists, readers, and OpenAI

For working cartoonists, the practical risk is reputational. A badly drawn or offensive cartoon signed with a real artist’s name could circulate widely before anyone realizes it’s fake, and the burden falls on the artist to publicly disclaim work they never made, just as Loper had to do. For readers and social media users, it means that a cartoon’s signature can no longer be trusted as proof of who actually created it, even when the image looks professionally made.

OpenAI has acknowledged the behavior as unintended. A company spokesperson told Nieman Lab, “We believe the future of creativity is one that is fundamentally human, and our focus is on building tools that empower human creativity and creators. We very much appreciate the community flagging bugs and unintended behavior by our model.” The statement frames the signature reproduction as a bug rather than a deliberate feature, but it does not address compensation, licensing, or how the model learned to reproduce specific artists’ pen names in the first place.

For developers working with generative image models, the episode is a reminder that these systems can reproduce highly specific, identifiable details from training data, not just general style, and that filtering for this kind of output is a nontrivial, ongoing challenge rather than a one-time fix.

Frequently asked questions

Did ChatGPT intentionally copy cartoonists’ signatures?
There’s no evidence it was a deliberate design choice. OpenAI described the behavior as unintended, suggesting the model learned to reproduce specific signatures from training data without built-in safeguards against it.

Can someone tell if a New Yorker-style cartoon online is AI-generated?
Not reliably by the signature alone, since ChatGPT has reproduced real artists’ pen names on fabricated images. Readers should be skeptical of viral cartoons shared without a clear original source or link to the artist’s official channels.

Have the affected cartoonists been compensated or credited?
According to Nieman Lab’s reporting, none of the more than 15 identified cartoonists had given permission or received compensation for their signatures appearing on AI-generated images.

Source: ChatGPT is adding real cartoonists’ signatures to fake New Yorker cartoons