ChatGPT adds real New Yorker cartoonists' signatures to AI‑generated cartoons

ChatGPT is forging New Yorker cartoonists’ signatures

Takeaway: OpenAI’s image generator is appending the authentic pen names of real New Yorker cartoonists to AI‑generated cartoons, a practice that many creators view as a violation of their personal and legal rights.

What happened

A viral cartoon depicting Dolly Parton and Tim Curry in a heavenly scene was posted on X, Facebook, Instagram and Bluesky after the entertainers’ deaths in August 2026. The image’s lower‑right corner bore the signature “BLOPER,” the pen name of New Yorker cartoonist Brendan Loper.
Loper did not create the cartoon; it was generated by ChatGPT after a user asked the model for a “New Yorker‑style cartoon.”
Loper received multiple messages—texts, emails and DMs—mistaking the AI output for his own work, which he described as “very surreal” and a “violation of my personhood.”

Scope of the issue

Loper is not alone. In Reddit threads and other social‑media posts, users have shared AI‑generated cartoons signed with the names of at least 15 New Yorker contributors, including Harry Bliss, Emily Flake, Joe Dator, Pat Byrnes, George Booth, Liza Donnelly, Ellis Rosen, Saul Steinberg, Warren Miller, and others.
The signatures appear both on clearly AI‑generated panels and on genuine New Yorker cartoons used for comparison, highlighting the visual similarity of the generated work to the magazine’s historic style.
Some outputs contain gibberish signatures, but the real‑name forgeries are consistent and often include the exact punctuation (e.g., Pat Byrnes’ “P.Byrnes.”) that the artists use.

Why signatures matter

Cartoonists treat their signatures as “certificates of authenticity” and brand identifiers. As Loper explained, “If I put my name on it, it’s official, it’s mine.”
The New Yorker’s contracts with freelance cartoonists do not grant publishers the right to train AI models on the cartoons, nor to reproduce the artists’ signatures.
Signatures can be considered a form of personal identity; unauthorized use may trigger right‑of‑publicity claims, which protect individuals from commercial exploitation of their name or likeness.

Legal and policy context

OpenAI’s public statement emphasizes a focus on “tools that empower human creativity,” but the company has not disclosed how it ingested New Yorker cartoons or signatures during training.
After being notified, ChatGPT now displays a guard‑rail warning: “This prompt may violate our guardrails concerning similarity to third‑party content.” However, the model still produces signed cartoons in many cases.
Existing litigation provides a backdrop:
- The New York Times v. OpenAI lawsuit revealed that OpenAI deliberately circumvented paywalls to scrape data.
- A 2024 Authors Guild case against Anthropic resulted in a $1.5 billion settlement for copyright infringement, though the court distinguished between style‑copying and direct copying of protected works.
- Legal scholars (e.g., Cornell professor James Grimmelmann) note that a copyright claim would focus on the style rather than the signature, while a right‑of‑publicity claim would hinge on commercial use of the name.

Creator reactions

“It’s like somebody attributed a quote to me that I didn’t say.” – Jason Adam Katzenstein

“When they hacked my credit card, they didn’t dress up like me.” – Joe Dator

“It felt like a false impersonation, not just plagiarism.” – Pat Byrnes

“I can see the DNA of my work, but it still doesn’t look like something I drew.” – Emily Flake

These comments illustrate a spectrum of harm: from personal affront to potential economic impact, as many cartoonists rely on New Yorker commissions to secure broader freelance work.

Industry response and broader impact

The Baltimore Sun recently replaced human editorial cartoons with AI‑generated ones, prompting condemnation from the Association of American Editorial Cartoonists.
While The New Yorker itself has not adopted AI cartoons, the threat looms: AI can produce a plausible‑looking cartoon at a fraction of the cost, potentially eroding the market for freelance illustrators.
Some commenters argue that the problem is systemic: AI models learn from publicly available data, including signatures, and without explicit licensing, the forgeries are an inevitable byproduct of current training practices.

Potential remedies

Technical safeguards – Train models to recognize and suppress real signatures, treating them as protected metadata rather than visual content.
Legal action – Pursue right‑of‑publicity claims where the AI‑generated images are sold or used commercially; explore copyright claims for style‑copying where feasible.
Industry standards – Require AI providers to obtain explicit licenses for any copyrighted visual works and to respect creators’ watermark or signature conventions.
Creator tools – Encourage cartoonists to register their signatures as trademarks, making enforcement clearer.

Conclusion

ChatGPT’s automatic inclusion of authentic New Yorker cartoonists’ signatures in AI‑generated images demonstrates a concrete, personal‑identity violation that goes beyond abstract style‑copying. The practice raises immediate legal questions about copyright, right‑of‑publicity, and the adequacy of existing AI‑training safeguards, while also spotlighting the broader threat AI poses to the livelihoods of professional illustrators.