Content creators are concerned about the use of their materials for AI training, following the launch of a government consultation that some say appears to favor AI firms.
The government says the consultation aims to improve legal clarity for the creative and AI sectors over how copyright protected materials are used in model training.
The proposals, lawmakers said, are intended to be fair, giving creators greater control over how their material is used by AI developers, enhancing their ability to be paid for its use, while giving AI developers wide access to material to train their models.
Similarly, the plans aim to provide greater transparency among AI firms over the data used to train AI models, as well as how AI-generated content is labelled.
"We are setting out a balanced package of proposals to address uncertainty about how copyright law applies to AI so we can drive continued growth in the AI sector and creative industries, which will help deliver on our mission of the highest sustained growth in the G7 as part of our Plan for Change," said secretary of State for science, innovation and technology Peter Kyle.
"This is all about partnership: balancing strong protections for creators while removing barriers to AI innovation; and working together across government and industry sectors to deliver this."
But balancing the needs of creators and AI firms is a tricky tightrope to walk, and many industry observers and creators are skeptical.
"The outcome will likely depend on getting the details right and whether ministers can get the balance right," said Matthew Sinclair, senior director of the Computer & Communications Industry Association (CCIA).
"In particular, it will be critical to ensure that the transparency requirements are realistic and do not ask AI developers to compromise their work by giving away trade secrets and highly sensitive information that could jeopardize the safety and security of their models."
Given long-running issues relating to the use of copyrighted materials in AI training, creators are concerned. Earlier this week, a series of publishers and author groups launched the Creative Rights In AI Coalition, calling for royalties to be paid to the creators of text, audio, or video used to train AI models.
"We are eager to see the development of a vibrant licensing market and support the sectors which rely on us for their future prosperity, but we can only do so with a robust copyright framework which preserves our exclusive rights to control our works and thereby act as a safeguard against misuse," the coalition said.
Amanda Brock, CEO at OpenUK, points out that the 10-week consultation and implementation period kicks any decision well into next year. Other countries, she added, have had clarity on fair-use exceptions for AI training for some time, and seen growth in their AI industries as a result.
Brock further warned that balancing the interests of creators and the AI industry is easier said than done.
"Any decision to require royalty or license payment for data usage as an outcome of the consultation inevitably disadvantages UK AI innovation," she said.
"The consultation outcome is predictable and the facts don’t change. There is direct conflict between the interests of these competing sectors."
Brock noted that licensing or royalty models are never going to be a long-term answer and merely represent a stopgap solution.
"In the same way as the linking agreements to connect web sites used in the early 2000s were soon made redundant by technology, content licensing is always going to be inhibitive of innovation in an AI world," she said.
"At best it will enable a few large companies with the wherewithal to enter licensing agreements to work with it."
