UK kicks off review into training AI models on copyrighted content
The UK government has initiated a review to regulate the use of copyrighted content by tech companies for training their AI models. The consultation aims to bring clarity to the creative industries and AI developers regarding the acquisition and utilization of intellectual property by AI firms for training purposes.
Concerns Raised
Some artists and publishers have expressed dissatisfaction with the unrestricted scraping of their content by companies like OpenAI and Google to train large language models. These models, including OpenAI's ChatGPT and Google's Gemini, are pivotal in today's generative AI systems.
Last year, The New York Times sued Microsoft and OpenAI, alleging copyright infringement and misuse of intellectual property in training their models.
Proposed Measures
The consultation will explore exceptions to copyright laws for AI training in commercial contexts while ensuring rights holders retain control over their content usage. It will also suggest ways for creators to license and receive compensation for the use of their content by AI developers.
The government emphasizes the need for collaboration between the creative industries and tech firms to establish effective and widely accepted standards for rights reservation and transparency.
Transparency Concerns
The government considers enforcing transparency from AI model makers regarding their training datasets and procurement methods to enable rights holders to track the usage of their content in AI training.
International Perspective
Previously, under former Prime Minister Rishi Sunak, the UK aimed to establish a voluntary AI copyright code of practice. Industry experts believe the UK is well-positioned to lead on intellectual property rights prioritization in AI.
Unlike the US, the UK is viewed as less influenced by lobbying from domestic AI giants, potentially allowing for more balanced legislation to protect individuals' intellectual property.
Emerging Challenges
The rise of multimodal AI, capable of processing images and videos alongside text, raises concerns about potential copyright violations by AI firms. OpenAI's recent release of Sora, an AI video generation tool, further underscores the need for robust regulations in this evolving landscape.
As the UK government delves into the complexities of AI model training on copyrighted content, the tech industry awaits the outcome of this pivotal consultation.
