Artificial intelligence companies OpenAI and Google have requested a government exemption to train their models on copyrighted material. The move comes as part of the US government’s AI Action Plan, which aims to enhance America’s position in the AI sector.
OpenAI believes that the federal government should provide an exemption to allow AI companies to learn from copyrighted materials without infringing on rights. This would help preserve American AI leadership and national security, while also securing Americans’ freedom to learn from AI.
In a blog post, OpenAI recommended maintaining tight export controls on AI chips to China and adopting AI tools broadly. The company has already begun offering a version of ChatGPT designed for US government use.
Google also published its recommendations for the AI Action Plan, stating that balanced copyright rules are essential for AI systems to learn from prior knowledge and publicly available data. The search giant argues that exceptions such as fair use and text-and-data mining allow for the use of copyrighted material without impacting rightsholders significantly.
OpenAI is currently facing lawsuits accusing it of copyright infringement, including ones involving major publishers like The New York Times.
Source: https://www.engadget.com/ai/openai-and-google-ask-for-a-government-exemption-to-train-their-ai-models-on-copyrighted-material-212906990.html