OpenAI and Microsoft are facing a lawsuit over alleged copyright infringement related to the training of their artificial intelligence (AI) models. The New York Times has filed a complaint, claiming that these companies used millions of its articles without permission or compensation to train chatbot models such as ChatGPT.
The Times argues that training generative AI models, like ChatGPT, with original articles from the newspaper constitutes copyright infringement. They also assert that OpenAI’s software replicated substantial portions of their work, including word-for-word replication in some cases. The newspaper estimates that this alleged infringement has resulted in the misappropriation of billions of dollars’ worth of its journalism. Furthermore, since Microsoft utilizes OpenAI’s technology and contributes to their research, the Times argues that Microsoft is also liable.
OpenAI and Microsoft are asserting a fair use defense, which is a legal doctrine allowing unlicensed use of copyright-protected works in certain circumstances. However, the Times believes the likelihood of a successful fair use defense is low in this case.
The first factor in assessing a fair use defense considers the purpose and character of the use. Here, the defendants’ purpose aligns with the Times’ goal of delivering news reporting to readers. However, this use lacks the transformative element typically required for a fair use argument. Additionally, given that AI software is sold for profit, the commercial nature of the use is evident.
The second factor focuses on the creativity of the underlying works. While AI treats the input work as factual data, the Times argues that their articles contain creative choices made during the presentation of factual material. Thus, the artistic and editorial aspects of the Times’ works remain significant.
The third factor revolves around the amount and substantiality of the copyrighted work used. The Times emphasizes substantial verbatim usage by OpenAI’s software, making it difficult for the defendants to assert that only a small portion of their works was used.
Lastly, the fourth factor considers the effect on the market for the copyrighted work. The Times contends that OpenAI and Microsoft’s use of their content negatively impacts its value, as readers can access the same information without paying for a Times’ subscription. This poses an additional challenge for the defendants in their fair use defense.
Companies and artists face challenges in safeguarding their work from being used for AI training. Some creators, like musician Grimes, have accepted the inevitability of AI using their work and are exploring licensing or selling options. The Times previously tried to negotiate an agreement with OpenAI for fair compensation in exchange for licensing their content.
To protect their work, writers, artists, and creators are encouraged to register copyrights. Registration guarantees access to remedies for infringement and notifies others of ownership. Consulting with an intellectual property attorney can help navigate how emerging technologies, like AI, impact proprietary rights.
In this case, as the dispute progresses to court, the question of whether AI training constitutes fair use or copyright infringement will be decided. The outcome will have implications for the use of copyrighted material in AI models and the responsibilities of developers and users alike.
