A Facebook User Says She Asked ChatGPT for “a New Yorker style cartoon.” Nieman Lab Says It Came Back Signed “BLOPER.”

A Facebook user wrote, in a comment on her own post, that she asked ChatGPT for “a New Yorker style cartoon” of Tim Curry and Dolly Parton at the Pearly Gates. Her description names no cartoonist. Nieman Lab reported on Oct. 5, 2026 that the image carries “BLOPER,” the pen name of New Yorker cartoonist Brendan Loper, who did not draw or sign it.

Nieman Lab reported on Oct. 5, 2026 that a viral cartoon of Tim Curry and Dolly Parton at the Pearly Gates carries the pen name “BLOPER,” the signature of New Yorker cartoonist Brendan Loper. Nieman’s Andrew Deck wrote: “But Loper didn’t draw the cartoon, or sign it.” A text from Loper’s brother came first, Nieman reported, and emails and direct messages from strangers followed. “I had people writing to me about this cartoon, asking if it was mine,” Loper told Nieman.

The Pearly Gates cartoon began, by its first poster’s account, as a request. A Facebook user wrote, in a comment on her Aug. 26, 2026 post, that she asked ChatGPT “to make me a New Yorker style cartoon of Tim Curry as Frankenfurter escorting Dolly through the Pearly Gates with her words here.” Her description names two entertainers and a scene, and no cartoonist. Nieman reports that ChatGPT generated the image.

From there it traveled: one tweet of it received 25,000 likes, Nieman reports, and other versions circulated on Facebook, Instagram and Bluesky. As viewed Oct. 11, 2026, neither the X screenshot Nieman published nor a Bluesky copy carries a caption saying who drew or signed it; in both, “BLOPER” sits in the bottom right corner.

Nieman Lab writes that a cartoon signature can “also act as a watermark that ties an illustration to its original creator.” Loper told Nieman: “The signature for me is the certificate of authenticity. I only put it on work that I’m putting on the internet that I feel needs to be recognized as from my hand.” And, of his name: “I’m not a territorial person, but my name is my name.”

In his own posts

Loper had written about the signature a month before Nieman did. “Yes, someone, some creep, is impersonating me, signing my name on terrible cartoons. That creep is ChatGPT,” he wrote on his Substack on Sept. 4, 2026. Two cartoons figure in that post: the Pearly Gates cartoon, which he learned of from a text from his brother at a wedding, and a second. Of the second, he says a stranger told him, earlier that year, that a ChatGPT request to “generate a new yorker cartoon that is very funny/clever” about therapy and mothers had produced an image bearing his name; the part of that request he quotes names no cartoonist either. That image, he wrote, “was never released to the wild.” He wrote that he “felt robbed of one of the most personal of personal things, my name.” On Oct. 6, 2026, on Instagram, he thanked Nieman Lab and Deck for an article about “OpenAI falsely signing AI-generated images with our names.”

Down to the period

Loper is not the only New Yorker cartoonist whose signature has appeared on ChatGPT cartoons, Nieman reports; dozens of images in online forums carry the signatures of other prominent contributors, including Harry Bliss, Emily Flake, Joe Dator, Pat Byrnes, Peter Vey and Jason Adam Katzenstein. In Nieman’s own tests, it reports, ChatGPT output cartoons with the signatures of George Booth, Liza Donnelly, Ellis Rosen and Saul Steinberg. Nieman documented more than 15 New Yorker cartoonists whose signatures “have been used by OpenAI’s image generator without permission or compensation,” partly in its own tests. Nieman reports that Byrnes, who published his first New Yorker cartoon in 1998 under the pen name “P.Byrnes,” has always signed that name with a period.

Several cartoonists Nieman spoke to, it reports, distinguished ChatGPT’s ability to mimic their illustration style from the reproduction of their signature. On our reading, a period after a name is not a style. “Across the more than a dozen ChatGPT cartoons I found signed with Byrnes’ name, every one ended with a period,” Deck wrote. Nieman does not print the prompt text behind the forum images or its own tests, beyond “New Yorker-style cartoons,” so its article cannot say whether any of those requests named a cartoonist. In Nieman’s own tests, it reports, not every image came with a signature, and some included pen names that were gibberish.

What the statement covers

An OpenAI spokesperson’s statement, as Nieman printed it, runs two sentences. The first opens with the company’s belief that “the future of creativity is one that is fundamentally human,” and goes on to its focus on tools for human creativity and creators. The second: “We very much appreciate the community flagging bugs and unintended behavior by our models, so we can address them.” Neither mentions signatures, cartoonists, Loper, The New Yorker or the August image. OpenAI declined to answer Nieman’s questions about how its models might have ingested signatures from The New Yorker’s archives during training, Nieman reports.

A rule that triggers on a living artist’s style

The one written artist rule this edition found is older. OpenAI’s system card addendum for GPT-4o native image generation, dated March 25, 2025, says the model “can generate images that resemble the aesthetics of some artists’ work when their name is used in the prompt.” In response to the concerns this raised in the creative community, the card says, OpenAI took “a conservative approach with this version of 4o image generation, as we learn more about how 4o image generation is used by the creative community.” It adds: “We added a refusal which triggers when a user attempts to generate an image in the style of a living artist.” As of Oct. 11, 2026, the card, read in full as text, never mentions signatures, The New Yorker or cartoons, and no source this edition read says which OpenAI model made the August image; Nieman says only that ChatGPT generated it.

OpenAI’s 2026 image documents, as this edition read them on Oct. 11, 2026, do not restate that rule. The ChatGPT Images 2.0 system card of April 21, 2026, the Images 2.5 card of Sept. 8, 2026, the Model Spec of Aug. 18, 2026, a help article and the launch posts state no rule on a living artist’s style, or on artists’ names or signatures. That shows what those documents say, not what OpenAI runs, and not that any rule was dropped. The Images 2.5 card describes policy checks at both ends, on the request before it reaches the image model and on the final image before the user sees it; neither, as described, names artists or signatures.

Nieman reports that after it notified OpenAI, ChatGPT began showing Deck a new message when asked for New Yorker-style cartoons: “This prompt may violate our guardrails concerning similarity to third-party content.” Nieman reports that message as ChatGPT’s output, not as a statement from OpenAI’s spokesperson, and as of Oct. 11, 2026 its wording appears in no OpenAI document this edition read. As of Nieman’s Oct. 5 story, Deck wrote, ChatGPT continued “to sign some of the generic cartoons it generates with the names of real New Yorker cartoonists.” On our reading, the one written rule we found keys on what a user asks for, and the message answered a request too; neither describes a check on a name the image itself carries. The name, Nieman reports, came back in the output.

What a signature is for

On our reading, a signature exists to answer one question: is this yours? On this cartoon the mark did not answer it. People wrote to Loper asking whether the cartoon was his.

Where the argument stops

In the sources this edition read, the request appears only as its first poster described it, not as a log of the prompt as sent.

As of Oct. 11, 2026, no source this edition read establishes why ChatGPT put real cartoonists’ names on its images. This edition did not re-test ChatGPT, and the Reddit forum, the X post and OpenAI’s full statement to Nieman went unread. Nieman found no evidence the cartoons are bought or sold, while reporting “signs that image generators are already displacing the work of professional cartoonists”; neither its article nor Loper’s posts, as this edition read them on Oct. 11, 2026, report a lost sale, commission or income from the signed cartoons.

We map the record, not any person, to Digital Derangement Syndrome™ (DDS™), our term for AI-mediated systems failing to recognize legitimate authority; here the signature itself, an authority mark, sits on work its owner did not make. Trust Transfer Failure (expertise never translated into machine-readable trust) is the closest fit. That mapping is an editorial judgment, not a source’s finding.