Rishi Sunak needs to decide whether to support Britain’s creative industries or bet everything on the artificial intelligence boom, Getty Images’ chief executive has said.
Craig Peters, who has led Image Library since 2019, made the comments amid growing anger in the creative and media sectors over the material being collected as “training data” for AI companies. His company is suing a number of AI image generators for copyright infringement in the UK and US.
“If you look at the UK, probably about 10% of GDP is made up of creative industries like film, music and television. I think it’s dangerous to make that trade-off. It’s a bit of a complicated trade-off to bet on AI, which is less than a quarter of the country’s GDP, much less than the creative industries.”
In 2023, the government, in response to consultation from the Intellectual Property Office, set a goal to “overcome the barriers currently faced by AI companies and users” when using copyrighted material, and promised to “support access to copyrighted works.” input to the model.”
This was already a step back from previous proposals for broad copyright exceptions for text and data mining. In a response to a House of Commons committee on Thursday, Viscount Camrose, a hereditary peer and under-secretary of state for artificial intelligence and intellectual property, said: This will help secure the UK’s place as a world leader in AI, while supporting the UK’s thriving creative sector.”
The role of copyrighted material in AI training is under increasing pressure. In the US, the New York Times sued OpenAI and Microsoft, the creators of ChatGPT, for using news articles as part of training data for their AI system. OpenAI said in a court filing that it is impossible to build an AI system without using copyrighted material.
Peters disagrees. Getty Images collaborated with Nvidia to create its own image generation AI that is trained using only licensed images.
The tide is changing within the industry as well. A dataset of pirated e-books, called Books3, is hosted by an AI group whose copyright takedown policy at one point even includes a costumed person pretending to masturbate with an imaginary penis while singing. Similar to the lawsuit by Getty and the New York Times, a number of other legal actions are underway against AI companies over potential training data breaches.
Ultimately, whether courts or even governments decide how to regulate the use of copyrighted material to train AI systems may not be the final word on this issue. Peters is optimistic that this result is not a foregone conclusion.
Source: www.theguardian.com
