THE DECODER

Singapore's government wants to use authors' works to train a large AI language model, but is facing resistance from writers. The government has requested permission to use published texts for its $52 million National Multimodal LLM Program (NMLP). But many authors and publishers are opposed, citing unresolved issues of compensation and copyright. They are concerned that AI could misuse their works. Without their consent, the NMLP would have to rely on lower quality content, which could negatively impact the model's performance. Authors around the world are increasingly resisting the unauthorized and unpaid use of their texts for AI training. Recently, eight major US publishers sued OpenAI and Microsoft for copyright infringement. OpenAI and other AI companies are currently entering into non-transparent agreements with select publishers to better position themselves for future AI models and ongoing litigation.