An anti-piracy group representing major Japanese video game developers like Square Enix, FromSoftware, and Bandai Namco has issued a warning to OpenAI, alleging that OpenAI may have committed copyright infringement. The claim is just one of many that have recently hit AI and LLM companies like OpenAI.
OpenAI is responsible for the creation of AI/LLM services like ChatGPT and the Sora 2 video generator. In recent times, criticism and concern have been voiced regarding the materials these services were trained upon, and whether OpenAI has the legal right to use them. Now, some Japanese game developers are taking notice that the Sora 2 video generator seems to be producing content that bears similarities to their own.
Major Companies Back New Bill to Kill Anime Piracy Sites in the US
Proposed bill would amend current copyright laws, with clear legal processes to issue blocking orders for foreign piracy sites in the US.
OpenAI Faces Copyright Infringement Allegations from Japan
An anti-piracy organization in Japan called the Content Overseas Distribution Association, or CODA, has made a written request to OpenAI to stop using its members' content to train the AI models. CODA states that it has "confirmed" that a "large portion" of the content produced by OpenAI's Sora 2 strongly resembles existing Japanese content, and that CODA considers the act of "replication during the machine learning process" to potentially be copyright infringement. CODA represents a number of Japanese media companies, including game developers Square Enix, Bandai Namco, and FromSoftware, as well as anime companies like Aniplex and Studio Ghibli. Studio Ghibli was notably the subject of a recent AI art trend that remade uploaded user images in the Studio Ghibli artstyle.
CODA has pointed out that OpenAI has an opt-out system, but that this doesn't mesh with Japanese copyright law. While the US has fair use rules, which often allow the use of existing content so long as it's used in a transformative manner, Japan does not. Instead, permission must be sought beforehand, which OpenAI seemingly did not seek before allegedly using copyrighted Japanese works as training material. As a result, CODA is specifically requesting that groups it's representing not have their content used for machine learning without permission, and requests a response from OpenAI regarding copyright infringement tied to Sora 2.
At the moment, OpenAI has not issued a public response to the claims. The company has a lot on its plate already, with the recent rollout of the Sora 2 video generator and major deals with companies like Amazon. However, it's also faced a lot of backlash, with publications like The New York Times, author groups, and George R. R. Martin suing or threatening to sue for similar claims. On the other hand, some game developers have already chosen to use AI within game development, such as EA.
Source: CODA