ChatGPT signs fake New Yorker-style cartoons with the names of real cartoonists

ChatGPT generates cartoons that imitate the style of The New Yorker and, in some of them, adds the signature of real cartoonists from the magazine who did not draw them. Andrew Deck documented it at Nieman Lab, a website that covers journalism.

The most widely seen case is a cartoon of Dolly Parton and Tim Curry that went viral in late August with the signature "BLOPER", the pen name of Brendan Loper. Loper did not draw it: ChatGPT had generated it after a user asked for "a New Yorker-style cartoon". Nieman Lab counted the signatures of more than 15 of the magazine's cartoonists, including Emily Flake, Harry Bliss and Joe Dator, used without permission or compensation.

After Nieman Lab notified OpenAI, ChatGPT began answering those requests with a warning that the prompt may violate its guardrails concerning similarity to third-party content. Even so, when the story was published it was still signing some cartoons with real names. OpenAI says it appreciates the community "flagging bugs and unintended behavior" by its models so it can address them, but it declined to explain how they learned those signatures.

Condé Nast, the publisher of The New Yorker, signed a licensing deal with OpenAI in 2024, but a spokesperson for the magazine says Condé Nast has never given an AI developer permission to train models on its cartoons. "My name is my name," Loper says. "It felt very much like a violation of my personhood."