Igeekphone, March 14 news, technology media MacRumors published a blog post yesterday (March 13), reporting that OpenAI has submitted an AI development proposal to the US government, calling for relaxed regulation, let AI companies freely use copyrighted materials, training AI models.
In its recommendation, OpenAI argues that the current plethora of legal restrictions imposed by states in the United States hinders the ability of AI companies to innovate and weakens the quality of training data available to American entrepreneurs.
OpenAI proposes a “copyright strategy” that advocates the free use of copyrighted data to train AI models, arguing that these models are “not copying works for public consumption” and comply with the “fair use” principle.
OpenAI noted that more than 781 AI-related bills have been introduced in states that impose compliance requirements on domestic companies and limit innovation, and the company recommended that the government provide “legal relief” to the private sector.
According to the blog post, OpenAI’s proposal has sparked complaints from creative workers such as artists, journalists, and writers, who believe their works are being used for “white whoring” to train AI models.
The New York Times has sued Microsoft and OpenAI, accusing them of using news articles to train AI models. In addition, image generation engines such as Dall-E and Midjourney have faced lawsuits over their use of millions of web images.