31 Jan 2024
In the last days of 2023, The New York Times sued OpenAI and Microsoft for copyright infringement, not because a feature published by The New York Times would ever be regurgitated from an input into ChatGPT, but because more than a million articles published by The Times have acted as the training tools for OpenAI’s artificial intelligence technologies.
Despite being such a major player in the publishing world, in the context of the data sets of OpenAI’s large language models (LLMs), The Times is a relatively small fragment of the resource that it draws on. What is significant is that The New York Times is the first major American media organisation to sue OpenAI and Microsoft over the unauthorised use of its intellectual property. A ruling in favour of the media giant, which is demanding the destruction of all LLMs trained on its copyrighted work, could bring about the end of OpenAI as we know it and fire the starting gun to countless other lawsuits in an overwhelming uprising in the defence of copyrighted material.
It had been OpenAI’s intention to head off this revolt by agreeing deals with key global players in the media industry and in December 2023, OpenAI had established permissions from Axel Springer, the German owner of brands including Politico, Business Insider, Bild and Welt, to use published content to train its LLMs. Dubbed as the ‘first of its kind’, the deal was quickly followed by similar arrangements with the Associated Press and Shutterstock.
The complexities of copyright have long been a shadow over the bright new dawn of AI, with artists of all different forms of media keeping a keen eye on how material is being generated, the inspirations AI content is being drawn from and the way it is being marketed, especially when there are clear influences and connections to established names in the industry.
Copyright used to be able to be linked directly to the origin of an idea or concept, but AI has broken that direct link with original thought. Instead, AI pulls together the thinking of millions of creatives to produce something new, but not truly original.
Are the wheels of the AI bandwagon about to fall off? In truth, almost certainly not. There are core functions and efficiencies that AI already has the capacity to deliver, providing businesses with quicker, cheaper and more accurate solutions – from coding and web development to data analysis and documentation comparison.
However, within these potential ‘quick wins’ for business, questions will continue to rage around security, authenticity and originality. Ultimately the battle lines of how far we allow the machines to take control are already being drawn.