In the past 24 hours, we have seen AI-generated images that represent Studio Ghibli versions. Elon Musk (19459035), ” The Lord of the Rings (19459035) “, President Donald Trump (). OpenAI CEO Sam Altman seems to have even a better grasp of the subject than President Donald Trump. His new profile picture is a Studio Ghibli style imagethat was presumably created with GPT-4o native image generator. ChatGPT users are uploading images and pictures and asking it to recreate them in new styles.
OpenAI’s latest update follows Google’s release of an AI image feature similar to its Gemini Flash model. This feature also caused a viral moment in March when users used it remove watermarks from pictures.
OpenAI and Google’s new tools make it easier to re-create copyrighted styles — just by typing a text request. The real concern is how these AI tools have been trained to mimic styles. Do these companies train on copyrighted work, and if they do, does this violate copyright laws?
This is the question at the heart of several ongoing lawsuits filed against generative AI models developers. Evan Brown, a lawyer with the law firm Neal & McDevitt and an expert in intellectual property, says that products like GPT-4o’s native image generator are currently operating in a gray area. Brown says that style is not explicitly protected under copyright. This means OpenAI’s generation of images that resemble Studio Ghibli films does not appear illegal.
Brown, however, says that it’s plausible for OpenAI to have achieved this likeness through training its model on millions frames from Ghibli films. Even if this was the case, courts are still deciding if training AI models to copyrighted works is protected by fair use. Brown said in an interview that “this raises the same questions we’ve asked ourselves for a few years now.” “What are the implications of copyright infringement when crawling the internet and copying into databases?”
The New York Times, along with several publishers, are suing OpenAI for allegedly training its AI models using copyrighted work without proper attribution or compensation. Similar claims have been made in lawsuits brought against other leading AI companies including Meta and AI-image-generation startup Midjourney.
An OpenAI spokesperson told TechCrunch that while ChatGPT does not allow it to copy “the style of living artists”OpenAI allows it to copy “broader studio styles”. Of course, there are still living artists credited with creating their studio’s unique style, like Studio Ghibli co-founder Hayao Miyazaki.
Users have also been able use GPT-4o’s native image-generation function to re-create the styles of other studios and artist. Someone else created a Marc Andreessen Portrait in the style Dr. Seuss (19459035) and a married pair recreated their Wedding photos in the style Pixar.
To test their ability to match Studio Ghibli’s style, we tested several popular AI Image Generators, including those available in Google’s Gemini and xAI’s Grok. OpenAI’s image generator was the most accurate in replicating the style of the animation studio.
OpenAI and Google’s latest image features are a huge leap forward for what AI models can produce, which is driving a surge of usage. OpenAI delayed its rollout of a new image tool for free-tier users Wednesday, citing a high demand. It may be that this is the most important thing these companies are doing today, but the courts will have to decide on their legality. Maxwell Zeff, a senior reporter for TechCrunch who specializes in AI and emerging technology, is available at
. Zeff covered the rise and fall of AI, as well as the Silicon Valley Bank Crisis, for Gizmodo and MSNBC. He is based out of San Francisco. When he is not reporting, he enjoys hiking, biking and exploring the Bay Area food scene.
View Bio
