Japanese Media Giants Take Measured Approach to OpenAI’s Sora Copyright Concerns
Table of Contents
A coalition of japan’s leading entertainment companies is addressing potential copyright infringements related to OpenAI’s new text-to-video model, sora, with a surprisingly cautious tone. The group, known as CODA, sent a letter to openai on October 28 outlining concerns about the similarity between sora 2’s generated content and popular Japanese media, including properties like Pokémon, video games from Hideo Kojima, and works from Studio Ghibli.
The emergence of Sora in late September sparked both excitement and alarm within the artificial intelligence community, as early outputs quickly revealed the potential for widespread copyright issues. The initial wave of generated videos included numerous instances of content closely resembling established intellectual property, prompting a swift response from rights holders globally.
A Different Approach to Copyright
CODA’s approach differs substantially from the more assertive legal strategies frequently enough employed by American copyright holders. Instead of issuing demands or threats of immediate legal action, the association has opted for a more diplomatic strategy. “The similarity between Sora 2 and Japanese images and video is the result of using Japanese content as machine learning data,” a CODA representative stated. The organization believes that the very act of replicating copyrighted material during the machine learning process may constitute infringement.
The letter to OpenAI outlines two key requests: that the company refrain from using content created by CODA members as training data without explicit permission, and that OpenAI provide a “sincere” response to any copyright complaints filed by CODA members. Notably, the organization has refrained from issuing ultimatums or setting strict deadlines.
This measured response is highly likely influenced by the nuances of Japan’s Copyright Act, specifically Article 30-4, wich addresses the use of copyrighted material in artificial intelligence growth. According to a government fact sheet, “exploitation for non-enjoyment purposes,” such as AI training and data analysis, “may, in principle, be allowed without the permission of the copyright holder.”
However, CODA emphasizes that this permissive environment does not negate the need for prior authorization. “Prior permission is generally required for the use of copyrighted works,and there is no system allowing one to avoid liability for infringement through subsequent objections,” the organization asserts. This suggests CODA is attempting to establish a precedent for proactive consent, even within a legal framework that might otherwise allow for
The situation highlights the complex challenges of applying existing copyright law to rapidly evolving AI technologies. As AI models become increasingly capable of generating original content, the lines between inspiration, imitation, and infringement are becoming increasingly blurred. CODA’s approach suggests a willingness to engage in dialog and establish clear guidelines,rather than instantly resorting to litigation. This strategy could prove pivotal in shaping the future of AI-driven content creation and copyright protection in Japan and beyond.
