TLDR
- Among GPT-4o's features is a striking AI image creator that's gained popularity for its impressive Studio Ghibli-like visuals.
- While the tool can emulate studios like Studio Ghibli, it brings up legal challenges related to copyright as the legality of AI training on copyrighted material is under scrutiny.
- A wave of lawsuits targets AI companies, including OpenAI, accusing them of copyright infringement for using works without obtaining permission.
- This latest AI image creator shines in text accuracy, faithfully executing prompts, and achieving stylistic congruence across a series of images.
- OpenAI's application produces remarkably faithful stylistic imitations, surpassing rivals like Google’s Gemini and xAI’s Grok.
OpenAI recently unveiled an innovative image creation tool as part of its GPT-4o suite. Flooding social media feeds, AI-crafted images that replicate Studio Ghibli’s unique style, associated with iconic Japanese films like 'My Neighbor Totoro' and 'Spirited Away,' have surged in popularity within hours of launch.
Within a mere day of its release, the tool has been used to create whimsical Ghibli-styled impressions of figures like Elon Musk, unique renditions of 'The Lord of the Rings,' and depictions of Donald Trump. Even OpenAI’s CEO, Sam Altman, seems to have adopted a Ghibli-inspired avatar, likely crafted via the new function.
The feature allows users to effortlessly transform existing images into diverse styles using ChatGPT, significantly contributing to its viral spread across platforms.
This feature emerges shortly after Google’s Gemini model introduced a similar capability, setting off its viral trend in March when users found ways to erase watermarks from images. Both tools simplify the recreation of copyrighted styles using straightforward text inputs.
The introduction of these AI art tools has revived discussions on copyright infringement. Legal actions against AI firms allege that these technologies are trained on copyrighted content without proper consent or compensation.
AI Art & Legal Gray Areas
Evan Brown, an intellectual property attorney at Neal & McDevitt, points out that these AI artistry systems navigate a controversial legal landscape. While the art style isn’t directly copyrighted, the method these systems utilize to learn them could stir concerns.
Brown highlighted a persistent inquiry about the implications of web-crawling and replicating external content in AI libraries, sparking debates about copyright infringement.
Major publications, such as The New York Times, join forces with other rights holders in active lawsuits against OpenAI, accusing the company of training AI models on copyrighted material without acknowledging the creators or offering compensation. Meta and Midjourney face parallel allegations.
OpenAI clarifies that ChatGPT refrains from mimicking the signature styles of living artists, although it does allow the replication of broader studio aesthetics, blurring lines when considering creators like Hayao Miyazaki, whose artistic vision set Studio Ghibli apart.
studio ghibli is out, dr seuss is in pic.twitter.com/4ECxwLLkoj
— Jordi Hays (@jordihays) March 26, 2025
Users have even adapted the tool to capture styles from other distinctive sources. Examples include whimsical sketches inspired by Dr. Seuss and wedding photos reimagined through Pixar’s imaginative lens.
By comparing the Stylus strokes and animation hues, tech journalists determined that OpenAI’s latest creation replicates Studio Ghibli’s enchanting essence with greater fidelity than rival offerings like Google’s Gemini, xAI’s Grok, and Playground.ai.
OpenAI asserts that this new visual creator balances 'aesthetic appeal with functionality.' It boasts about its tool's prowess in precise text rendering, genuine adherence to user directives, and leveraging the expansive GPT-4o knowledge base.
One standout feature is the generator’s capacity to uphold stylistic integrity across multiple visuals in a conversational sequence, facilitating smoother iterative refinement while preserving coherence in elements like character design.
Designed with advanced “instruction adherence” capabilities, the model deftly processes complex prompts featuring up to 10-20 distinct items while maintaining their interrelations, enhancing control over prior iterations that struggled beyond handling 5-8 elements.
OpenAI has incorporated safeguards into its new feature, with all media containing C2PA metadata to indicate AI creation. The system also filters content violating their guidelines, imposing stricter rules when real individuals are involved.
ChatGPT users, ranging from Plus, Pro, Team, to free-tier members, now have access to this generator by default, with expansions to Enterprise and Education sectors anticipated shortly. Developers will gain API access to leverage GPT-4o for image generation soon.
While acknowledging certain challenges like occasional image trimming issues, hurdles in multilingual text precision, and specified editing tasks, OpenAI commits to refining these aspects in forthcoming tools updates.
The burgeoning demand for this breakthrough image feature pressured OpenAI to stagger its free-tier user rollout, attributing the delay to unexpectedly high traffic volumes.
This technological leap signifies a significant stride in AI's creative potential, yet court determinations remain awaited to resolve whether these models infringe upon existing copyright statutes.
