ChatGPT generates fake New Yorker cartoons, complete with artist signatures

OpenAI's ChatGPT has sparked controversy again, this time for its ability to produce counterfeit New Yorker cartoons featuring the signatures of real artists. The issue has been brought to light by the viral spread of seemingly similar images on social media platforms, according to Nieman Lab.

The chatbot's success in creating these forged images is attributed to two primary factors: the widespread popularity of New Yorker cartoons online and the process of internet-wide scraping designed to gather content for model training. Multiple publishers, such as Condé Nast, which owns the New Yorker, have been advocating for federal legislation to curb unauthorized scraping of their content. Condé Nast previously entered into a partnership with OpenAI to compensate for the loss of revenue due to unlicensed data usage.

Despite this, a New Yorker spokesperson made it clear that the agreement did not authorize OpenAI to train its models on cartoons or utilize its logo and artwork in image generation. Furthermore, the publication emphasized that the contributing cartoonists retain full copyright over their work.

OpenAI addressed the plagiarism allegations, stating their commitment to fostering human creativity through their tools. They welcome community feedback regarding any issues or mistakes in their models' behavior to rectify them. Currently, ChatGPT provides a warning to users when attempting to generate New Yorker-style images, stating that the prompt may violate guidelines concerning similarity to third-party content.

However, the chatbot continues to generate images in the desired style, including the inclusion of signatures, despite these warnings.