A Controversy Unfolded
Artificial Intelligence has brought about a significant transformation across various industries globally. However, as AI capability accelerates, it dances on the fine line of copyright law – a problem faced by OpenAi’s popular AI-driven app, Sora 2. Recently, it found itself on the receiving end of a wave of criticism from the Content Overseas Distribution Association (CODA), an organization that represents Japanese intellectual property holders like Studio Ghibli and Bandai Namco.
Last week, CODA released a letter directed to OpenAI, demanding an immediate halt to the use of their members’ content to train OpenAI’s app, Sora 2. According to CODA, using copyrighted content during the machine learning process could constitute copyright infringement. The artificial intelligence model of Sora 2 creates content featuring characters protected by copyright law. This, they feel, is a breach of their intellectual rights.
Japanese IP Overuse
Started on the 30th of September, Sora 2 has significantly overused Japanese intellectual property (IP). It generated an incredible amount of content saturated with Japanese characters and artwork, leading the Japanese government to officially request OpenAI to cease replicating Japanese creations. This is not the first occasion on which OpenAI has heavily borrowed from Japanese content. Back in March, during GPT-4o’s launch, the highlight was an array of images produced in the much-loved “Ghibli-style”. A point of interest is that even Sam Altman’s profile photo on X resembles a style typical of Studio Ghibli.
Revision of Policy
OpenAI’s President, Sam Altman, announced last month that a change in Sora’s policy is imminent. This considered, CODA retorted by stating, using an opt-out policy at all could have been a violation of Japanese copyright law. CODA points out that under Japan’s copyright regulations, any use of copyrighted work generally requires prior permission. There isn’t a system that provides for infringement liability avoidance through subsequent objections.
While recognizing the rapid advancement in AI and its overwhelming potential, CODA insists that intellectual property must be respected. They have asked OpenAI to respond sincerely towards their members’ copyright claims and stop employing their content for machine learning without explicit permission. This includes not only the output from Sora, but also Japanese IP being used as training data.
As this story unfolds, the debate surrounding AI and copyright laws presents a complex challenge, needing balance between intellectual property rights and technological progression. OpenAI’s next move will undoubtedly set a crucial precedent in this key debate.
Originally published at The Verge.
