OpenAI’s ChatGPT can’t seem to shake its plagiarism problem.
Amid an ongoing battle with authors and publishers over its text-based outputs and continued criticism that the chatbot’s creative mind is ripping off human artists, the chatbot is under fire again. This time because of poor imitations of New Yorker cartoons, complete with signatures of real, paid artists.
Recent viral outputs shared on social media seem to show a repeated pattern of the chatbot adding fake signatures of prominent New Yorker staff cartoonists to its artificially-generated images, reported Nieman Lab.
ChatGPT was able to generate the signatures of more than a dozen professional cartoonist and affix them to the fake images, often rendered in the signature style of the publication, Nieman Lab reported. The artists were not asked permission or provided compensation.
“I can see the DNA of my work in there, but it still doesn’t look like something I drew. It’s kind of like seeing a distant cousin,” contributor Emily Flake told Nieman Lab.
Mashable Light Speed
The New Yorker signatures are likely the result of two factors: The ubiquity and popularity of New Yorker cartoons across the internet and internet-wide scraping that aims to siphon content for model training.
Many publishers, including Condé Nast, are now pushing Congress to enact federal laws that would prevent unlicensed bot scraping. In recent years, these same publishers have signed deals with AI companies to recoup revenue lost from previously unlicensed data grabs, including a 2024 partnership between publisher Condé Nast and OpenAI. Under the deal, OpenAI paid for use of content from Condé Nast brands, including New Yorker, Vogue, GQ, Vanity Fair, and Bon Appétit.
However, a New Yorker spokesperson clarified that the deal did not grant OpenAI permission to train its models on cartoons or use its logo and art in image generation. In addition, the publication clarified that contributing cartoonists retain copyright of their work.
In response to reports of potential plagiarism, an OpenAI spokesperson said: “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 models, so we can address them.”
The chatbot now warns users of potential content violations when prompted to generate New Yorker-style images. It reads, “This prompt may violate our guardrails concerning similarity to third-party content.” However, ChatGPT will still generate images in the style, including signatures, the publication reported.
