Chatgpt and Ghibli images, it’s over
Chatgpt hits the screw on the generation of certain images © OPENAI – Chatgpt
CAP change for chatgpt. Openai’s AI seems to want to end the trend to generate images that are too inspired by certain creators. After first measures put in place at the end of March to avoid abuses, the Sam Altman firm seems to have made a radical decision to avoid attracting the wrath of copyright.
End clap on the generation of Ghibli et Cie images with chatgpt
While it was still possible to create images this morning in the style of Ghibli, Wes Anderson or any other personality, Openai seems to want to put an end to this practice. For a few hours, Chatgpt has refused to submit to the exercise, evoking an offense on the content of Openai content.
Chatgpt refuses to generate images in Ghibli style © Chatgpt – screenshot
When you ask the AI why it was possible to do this this morning, but no longer now, the chatbot explains that it can be linked to a recent change or update.
Excellent question – and I understand your frustration. The difference is likely to come from a temporary change or a recent update of image generation rules, which can be adjusted automatically depending on the context, the type of content requested or other security parameters.
According to the tests we have carried out, only GPT-4O currently seems to apply this new limitation. The recently deployed models: O3, O4-Mini and O4-Mini-High would not have yet benefited from this update. In addition, this copyright linked to the moment only seems active on the computer version.
All of these elements suggest that new restrictions are gradually being deployed in Chatgpt in terms of the generation of images that can violate copyright.
