Generative AI has become a hotbed for copyright disputes, with recent lawsuits and debates highlighting the challenges that arise when artificial intelligence models are trained on copyrighted works without the creators’ consent. The Authors Guild, along with prominent authors such as George R.R. Martin, recently filed a lawsuit against OpenAI, alleging that the company’s chatbot, ChatGPT, was trained on their literary works without authorization.
Key Takeaway
Generative AI has sparked copyright controversies, with lawsuits and debates surrounding unauthorized use of copyrighted works. Startups are grappling with how to address training models on artists’ works, and the legal complexity of fair use doctrine adds further challenges. Some companies have introduced opt-out mechanisms, while others have established funds to share revenue with artists. The copyrighting of AI-generated works remains an open question, but companies like Runway stand behind creators and their content.
The Challenges Faced by Generative AI Startups
OpenAI is not the only company grappling with these challenges. During the Disrupt 2023 conference, Anastasis Germanidis, co-founder of Runway, a generative AI tools developer for video, acknowledged that his company is still in the process of determining the best approach to training AI models on artists’ works. Runway, like many other generative AI startups, does not disclose the specific sources of their training data, which raises the question of potential copyright infringement.
Other companies in the generative AI space, such as Stability AI, Midjourney, and DeviantArt, have also faced legal action from artists who claim that their copyrighted works were used to train AI models without permission. Getty Images even filed a lawsuit against Stability AI in the U.K., accusing the company of unauthorized copying and processing of millions of images.
The Legal Complexity and Fair Use Arguments
Generative AI companies often argue that they are protected by the fair use doctrine in the United States. However, the matter is far from being resolved, and it is unlikely to be settled in the near future. Copyright laws and court decisions have not provided a clear path on this issue.
In an attempt to protect themselves from potential legal challenges, some generative AI vendors, such as Stability AI, have implemented opt-out mechanisms, allowing artists to exclude their works from model training. OpenAI recently established a channel for artists to indicate that their artwork should not be used for future model training. Additionally, some platforms have established communal funds to share revenue generated by generative models with the artists whose works were used for training.
However, Runway currently does not offer an opt-out option or a contributor fund. Nevertheless, Germanidis hinted that the company is actively considering these measures to better support artists using their platform.
Copyrighting AI-Generated Works
While the focus has mostly been on protecting the copyrights of human creators, there is also an open question regarding whether AI-generated works can be copyrighted. The U.S. Copyright Office has recently solicited comments on the subject, and court decisions have not yet provided substantial clarity.
Germanidis, on behalf of Runway, claimed that content generated through their platform can be copyrighted. While he stopped short of promising a policy like Microsoft, which assumes liability for copyright-related legal damages incurred by customers using their AI services, Germanidis assured that Runway would provide defense and support to their customers when needed. He emphasized the company’s commitment to empowering artists and ensuring that the content they create belongs to them.
