Adapted by:
Javier Pastor
Senior WriterComputer scientist turned tech journalist. I've written about almost everything related to technology, but I specialize in hardware, operating systems and cryptocurrencies. I like writing about tech so much that I do it both for Xataka and Incognitosis, my personal blog.
Karen Alfaro
WriterCommunications professional with a decade of experience as a copywriter, proofreader, and editor. As a travel and science journalist, I've collaborated with several print and digital outlets around the world. I'm passionate about culture, music, food, history, and innovative technologies.
The viral trend of ChatGPT-generated images mimicking Studio Ghibli’s iconic style has sparked an interesting response from OpenAI—one aimed at preventing bigger problems down the road.
Watermarking. OpenAI is preparing to add watermarks to images created with its new AI model, ImageGen. According to Bleeping Computer, researcher Tibor Blaho discovered evidence of this feature in the ChatGPT Android app.
Text has been watermarked before. OpenAI previously explored watermarking systems for AI-generated text. Now, it appears the company is developing a similar solution for its image generator—likely to become a popular feature among new users.
A promising idea. Several companies have been working to develop standardized watermarks for AI-generated images. One such effort is the C2PA standard, supported by OpenAI and others, which continues to gain traction. The recent discovery in ChatGPT’s Android app suggests OpenAI plans to implement this feature natively for all generated images.
Pay to skip the watermark? Bleeping Computer also reported that sources close to OpenAI described two types of AI-generated images: Watermarks would apply to those created by free-tier users, while ChatGPT Plus subscribers could save their images without watermarks. This model resembles image banks, where users pay to access watermark-free photos.
A safeguard against lawsuits. OpenAI’s tool has gone viral for its ability to generate artwork that closely resembles Studio Ghibli’s animation. But that popularity also brings legal risks. Copyright remains a gray area for AI-generated content, and watermarking may help identify the source of such images—potentially avoiding lawsuits.
The fair use vs. copyright debate. In an appendix outlining risks associated with GPT-4o and its image model, OpenAI acknowledged the tool can generate images “that resemble the aesthetics of some artists’ work when their name is used in the prompt.” The company, like others in the space, argues this falls under “fair use” for training and generation. However, the line between fair use and copyright infringement continues to blur, raising concerns for artists, platforms, and users alike.
Image | Xataka On with ChatGPT
Log in to leave a comment