Ghibli’s fever has forced Openai to adopt an unprecedented measure: add water mark

He Viral phenomenon of the images generated by chatgpt that I mimic those of Studio Ghibli It has apparently caused an interesting reaction from OpenAI. One with which to avoid major evils.

Water marks. The company is preparing the inclusion of water brands in the images it generates Your new AI modelImagegen. As indicated in Bleeping Computer, a researcher named Tibor Blaho He has found References to these water marks in the Android application of Chatgpt.

They already did it with texts generated by AI. OpenAi has already raised long ago Your own system to integrate water marks into the texts generated by AI. Now it seems to be preparing something similar for its deployment in an image generator that is precisely becoming a successful tool for Capture a lot of new users.

An idea with a lot of future. We have been talking about the efforts from various companies to create some type of standardized water brand. He C2PA standardthat among other things has been supported by OpenAIit is the one that is gradually gaining ground. This discovery in the Android of Chatgpt reveals that intention to implement it natively by generating any image from the OpenAi chatbot.

Pay to avoid water brand? In Bleeping Computer they indicate how sources close to OpenAi talk about two types of images of AI: Water marks will be in the images generated with grauite accounts, but chatgpt plus subscribers can save them without water mark. It is something similar to what happens with image banks that have free images without water marks and others that do and that force to pay to remove them.

A measure to avoid demands. The ability to create images that imitate those of Studio Chibli has unleashed a viral fever for this Openai tool, but that also raises potential judicial actions. Copyright is still a controversial element of everything that surrounds AI models, and water brands would at least serve to it conform their origin and avoid legal demands.

The debate on fair use and copyright. In Your appendix On risks of use of GPT-4O and its image of the images, OpenAi mentions that the tool can be used to create images “that look like the aesthetics of some artists.” She and others They grab the concept of “fair use” of those tools when training them and generating images. However, the thin line that separates that fair use from a Copyright violation It is increasingly tense, and the risk is clear.

Image | Xataka with chatgpt

In Xataka | The price to be paid for having ia is the looting of all the internet content. And perplexity is just the last example