Summary
-
OPENAI’s GPT-4O model allows users to generate images and restore photos, but may soon be added waterproof for free level users.
-
The media generated by the AI filigree are a hot topic, but Openai tests the filigrane only for the images created by free level users.
-
The GPT-4O version of Chatgpt offers a generation of images, but the watermark can be used to encourage users to subscribe to the premium levels.
Openai has focused its efforts on popularization Chatbot experience Chatgpt offered since the initial public launch. At the time, the generation of images was managed by specialized AI tools like Midjourney. A year later, the limits are blurred because chatbots can also create simple images. Meta Ai and Google’s Gemini do it, but Openai is rather new on the stage with its GPT 4O model. Now the company would test the filigree for the images generated.
Related
GPT-4O was released to the public Free last week, after a short period when only paid subscribers could access it. In addition to several other improvements, this new model unlocks the image generation for users, you can provide a fully textual prompt or use an existing image and make it connect. The Internet is in excitement with Chatgpt Restyling Photos to also match the artistic style of Ghibli studio. However, to limit the assessment of an increase in use, free users are likely to generate only two images per day.
However, the creeping use of the art style of the Ghibli studio seems to have highlighted another problem with the GPT -4O – the lack of watermark. The IA researcher Tibor Blaho, who passes @ btibor91 on x, Recently shared as version 1.2025.0912509108 of the Android Chatgpt application mentions a new watermark imagegen (via Bleeping Compompute). Imagegen is the model that fuels the capabilities of generation of images of the multimodel experience that GPT-4O is.
Posters only on free level production?
A strange choice
While the AI industry collectively makes a watermark for the media generated by Ai-Ai, it is therefore never usable as a substitute for the content created by man, Openai does something bizarre by testing the watermark only for images generated by free level users. This might look like how Stock Watermark image distributors are free download copies and request payment for greater resolution or waterproof results, serving lure for paid chatgpt levels.
This is probably distinct from the digital watermark which distinguishes human work from the production of AI, but it will be interesting to see how users take the change when it deploys. Openai has not yet mentioned this publicly or engaged in a chronology, but rivals like Gemini and Meta AI already added filigranes visible to the images generated.