AI companies are facing mounting pressure over the use of copyrighted material to train their advanced models. OpenAI, the developer behind the groundbreaking chatbot ChatGPT, has stated that creating tools like ChatGPT would be impossible without access to copyrighted data. The company has been embroiled in a lawsuit with the New York Times and Microsoft, accusing them of “unlawful use” of copyrighted work to create their products.
In a submission to the House of Lords, OpenAI emphasized that copyright laws affect almost every form of human expression, making it necessary to use copyrighted materials when training AI models. OpenAI argued that limiting training data to out-of-copyright books and drawings would result in inadequate AI systems that do not meet the needs of today’s citizens.
The defense of AI companies hinges on the legal doctrine of “fair use,” which allows for the use of copyrighted content under certain circumstances without seeking permission from the owner. OpenAI maintains that copyright law does not prohibit training their models.
This issue extends beyond OpenAI. Getty Images, for instance, is suing Stability AI, the creator of Stable Diffusion, for alleged copyright breaches. In the US, music publishers including Universal Music are suing Anthropic, the Amazon-backed company behind the Claude chatbot, alleging misuse of copyrighted song lyrics for training their model.
Amid concerns about AI safety, OpenAI has expressed support for independent analysis of its security measures. The company endorses “red-teaming,” whereby third-party researchers emulate the behavior of rogue actors to test the safety of AI systems.
AI companies, including OpenAI, have also agreed to collaborate with governments on safety testing their most powerful models before and after deployment, following an agreement reached at a global safety summit in the UK last year.
The source of the article is from the blog newyorkpostgazette.com