The methods used by tech companies to train the majority of generative AI models are likely to be “illegal”, according to a senior executive who stepped down from British startup Stability.
Ed Newton-Rex, who recently resigned as VP of audio at London-based AI unicorn Stability over ethical concerns, told UKTN that the “jury is still out” on whether companies like OpenAI and Google are breaking copyright law by training their models on creators’ data.
Prior to his resignation, Newton-Rex was running Stability AI’s audio department, which was working on a music generation tool. He left Stability because he did not agree with “the company’s opinion that training generative AI models on copyrighted works is “fair use’”.
It comes as rapid advances in generative AI are pushing the limits of copyright law. Intellectual property lawyers say most existing copyright laws were not written with AI in mind....