AI and Copyright: OpenAI's Ghibli-Style Video Sparks Debate

AI Image Generation and Copyright Concerns

Barely a day has passed since ChatGPT’s new AI image generator became available, yet social media is already showcasing a wealth of AI-created content. Many of these images mimic the distinctive style of Studio Ghibli, the renowned Japanese animation studio responsible for acclaimed films like “My Neighbor Totoro” and “Spirited Away.”

Viral Trends and User Experimentation

Within the last 24 hours, AI-generated depictions of Studio Ghibli-style Elon Musk, scenes from “The Lord of the Rings”, and even former President Donald Trump have surfaced. OpenAI’s CEO, Sam Altman, has seemingly adopted a Studio Ghibli-inspired image for his profile picture, likely created using GPT-4o’s integrated image generator.

Users are actively uploading images and requesting ChatGPT to reimagine them in various artistic styles.

Competition and Similar Features

This update from OpenAI follows Google’s introduction of a comparable AI image feature within its Gemini Flash model. This feature previously gained attention in March when it was utilized to remove watermarks from images.

Both OpenAI and Google’s latest tools simplify the replication of copyrighted work styles through text prompts. However, a significant question arises regarding how these AI tools are trained to emulate such styles.

The Legal Landscape of AI Training

Are these companies utilizing copyrighted materials during the training process, and if so, does this constitute a violation of copyright law? This is the central issue in several ongoing legal battles against generative AI model developers.

Navigating the Legal Gray Area

Evan Brown, an intellectual property lawyer at Neal & McDevitt, explains that current legal precedents place these products in a gray area. Copyright law does not explicitly protect artistic style itself, suggesting OpenAI isn’t immediately infringing by generating images resembling Studio Ghibli films.

However, Brown posits that OpenAI may have achieved this stylistic likeness by training its model on a vast collection of frames from Ghibli’s films. Even if this is the case, courts are still deliberating whether training AI on copyrighted works qualifies as fair use.

Concerns About Web Crawling and Data Usage

“This situation echoes the questions we’ve been grappling with for the past couple of years,” Brown stated in an interview. “What are the copyright implications of systematically crawling the web and incorporating data into these databases?”

Ongoing Lawsuits and Claims

The New York Times and numerous publishers are currently pursuing legal action against OpenAI, alleging that the company trained its AI models on copyrighted content without appropriate attribution or compensation. Similar claims have been made in lawsuits against other prominent AI companies, including Meta and Midjourney.

In a statement provided to TechCrunch, an OpenAI spokesperson clarified that ChatGPT is programmed to avoid replicating the style of individual living artists. However, the company permits replication of “broader studio styles.” It’s important to note that living artists often play a pivotal role in defining their studio’s unique aesthetic, such as Hayao Miyazaki, co-founder of Studio Ghibli.

Expanding Style Replication

Users have demonstrated the ability to recreate styles from other studios and artists using GPT-4o’s image generation capabilities. Examples include a portrait of Marc Andreessen in the style of Dr. Seuss and a couple’s wedding photos reimagined in the style of Pixar.

Comparative Performance

We evaluated several AI image generators – including those found in Google’s Gemini, xAI’s Grok, and Playground.ai – to assess their ability to emulate Studio Ghibli’s style. Our testing revealed that OpenAI’s new image generator produced the most accurate replication of the animation studio’s aesthetic.

For the time being, the new image features from OpenAI and Google represent a significant advancement in AI-generated content, driving increased usage. OpenAI postponed the release of its new image tool to free-tier users on Wednesday due to overwhelming demand.

While increased usage is a key metric for these companies currently, the ultimate determination of their legality will rest with the courts.