– Guest article, written by Vincent Lautier –
For a few hours, you have not been able to miss, Chatgpt has struck hard with a new feature that allows you to generate high -quality images. Korben has also tested all this on this articleand it’s really very clean as results.
One of the most interesting possibilities is the possibility of applying precise visual styles like that of the Ghibli studio. Result ? Social networks are invaded by personal photos, memes or historical scenes revisited in a style reminiscent of the universe of Hayao Miyazaki. The visual quality of the images produced is crazy, but the question of the legality of these creations is much less clear.
Visually stunning results
It must be recognized that the tool is effective. In a few seconds, Chatgpt transforms any cliché into an illustration worthy of a Ghibli film. The round faces, the soft colors and the simple outfits immediately recall the aesthetics of masterpieces as My neighbor Totoro Or Chihiro’s journey. As you can see below with this photo of Korben and I, it’s really super clean.
But the craze goes beyond simple personalization. Many creations shared on social networks also reinterpret iconic images: from the Trump-Zelensky meeting in Ghibli mode to memes diverted in Japanese animation. Openai does not seem against this appropriation as long as it does not touch certain protected works like Disney or Dragon Ball.
In fact, at first there was no limits, but the screw was a little tight, and generating visuals inspired by Pokemon for example, begins to be more difficult. On the other hand, no problem generating photos in the style of Naruto or the Simpsons.
I have also just tested with a photo taken by me during a shoot (that you can see here), frankly the result is great, I want to test on all my photos! On the other hand there are many limitations, typically it blocks direct if there is, if only a little nudity, and so much the better to avoid problematic deepfakes.
A gray legal zone
In fact where it gets stuck, it’s really legal. Because if Openai assures that his model does not train on protected works of living artists, all this still lacks clarity. Evan Brown, a lawyer specializing in intellectual property, recalls that the visual style of a work is not protected by copyright, but the way in which Chatgpt succeeds in imitating it poses question. Basically, did Optai used millions of images of Ghibli films to train her AI? And if so, is it legal?
In reality, the problem is really not new. OPENAI, like other AI companies, is the subject of several legal actions for having trained its models on protected works. And France is no exception: in March 2025, authors’ organizations filed a complaint against Meta for similar practices.
Rules to be defined
For the moment, legislation is late on these questions. On the one hand, Openai welcomes it to be able to offer a functionality that amuses millions of users (me first). On the other, the artistic community is worried about seeing their style recovered without any control or remuneration.
The real question is therefore simple: should copyright be completely reviewed in the era of artificial intelligence? Because at this rate, creative AI may blow up many legal locks.
Small precision in passing, at first the functionality was open to all users of Chatgpteven free. In front of the mass of users, Openai decided a few hours ago to limit the functionality to Chatgpt Plus subscribers and beyond. So if you want to test now right now, you will need a paying subscription.
Article published by myself, Vincent Lautierinvited by friend Korben. You can also Follow me on Blueskyor read The unlikely little tests that I sometimes publish in the “Tech Gadgets” category !
Source link
Subscribe to our email newsletter to get the latest posts delivered right to your email.
Comments