Publishers, artists and media organisations are lining up to oppose potential UK rules that would formally authorise AI companies to mine their output without permission and for free.
Content creators agree that legislation may be needed to correct an existing free-for-all in which tech firms plunder the internet for the vast quantities of data needed to train their artificial intelligence models.
They fear their talents are being exploited to develop technologies that could ultimately replace them.
The government, meanwhile, is torn between its desire not to alienate either the tech companies it hopes will help turn the UK into an AI powerhouse or a creative sector that punches above its weight in a struggling economy.
Prime Minister Keir Starmer attempted to take the middle road this week in a letter to news publishers in which he recognised the basic principle that publishers should seek compensation for the use of their content by AI companies.
“Both artificial intelligence and the creative industries - which includes news and media - are central to this government’s driving mission on economic growth,” he told the News Media Association.
The erroneous belief that content is free
One option reported as under consideration by the government is to allow AI companies to ‘scrape’ online content unless its creators actively opt out. That, however, would put the onus on the creators to assert their copyright, or at least to try to monetise it via fees from the AI developers.
It is an option that might not be a major challenge for a big media outfit, but it would be a further burden on small organisations and individual creatives. They might not even be aware that their work is being poached or who is doing the poaching.
They might be better served by a system that required tech companies to invite content producers to opt in to having their data mined in exchange for suitable recognition and compensation. But that would put an additional burden on AI developers that might deter them from investing in the UK.
The debate underlines a great fallacy of the connected internet age: the erroneous belief that content is free
The debate underlines a great fallacy of the connected internet age: the erroneous belief that content is free. It is one apparently shared by Mustafa Suleyman, the British tech entrepreneur who heads Microsoft AI.
In an interview this year, he argued that anything published on the web is up for grabs to train AI models because it is essentially ‘freeware’, a principal he described as being established at the dawn of the internet age in the 1990s.
That was in the idealistic days when the World Wide Web was conceived as a global interchange for the free sharing of knowledge rather than as a licence for tech billionaires to print money.
Nowadays, as a UK House of Lords communications and digital committee pointed out in February: “Some tech firms are using copyrighted material without permission, reaping vast financial rewards.”
Dependence on copyrighted material
The success of AI companies is entirely dependent on the amount of existing data, mostly gleaned from the open internet, that it feeds into its machines.
In a submission to the same House of Lords committee, the developer OpenAI acknowledged its dependence on copyrighted material to create tools like its ChatGPT chatbot.
“Because copyright today covers virtually every sort of human expression - including blogposts, photographs, forum posts, scraps of software code, and government documents - it would be impossible to train today’s leading AI models without using copyrighted materials,” according to OpenAI.
It, like other developers, takes the position that the way it uses data to train its AI models already complies with copyright laws. They like to compare their machines to an author or student consulting a library in pursuit of academic research.
An executive at London-based Stability AI quit over the tech firm’s view that it was legtimate to use copyrighted work without permission to train its products
Not everyone, even within the industry, agrees. Ed Newton-Rex, an executive at London-based Stability AI quit over the tech firm’s view that it was legtimate to use copyrighted work without permission to train its products.
He has since organised a public statement asserting that unlicensed use of creative works for training generative AI is a major, unjust threat to the livelihoods of their creators. It has been signed by more than 30,000 international authors, musicians, film actors and others.
How to provide confidence to innovators and investors?
The previous Conservative government attempted to resolve the tech v. creatives impasse by sponsoring a consultation between the two sides. But it conceded defeat in February when it admitted there was no prospect of the two sides agreeing on an effective voluntary code.
If Starmer’s government fails to break the deadlock in further consultations, it would have to opt for legislation.
Google warned that the UK risked being left behind in the global race to develop AI unless it set out a comprehensive agenda
In the meantime, large-scale content producers are taking their own initiatives to protect their intellectual property. The UK’s Financial Times is among a number of international media companies that have signed deals to licence their archived content to AI companies.
Under the Financial Times’ agreement with OpenAI, the ChatGPT developer is allowed to use its content to provide text, images and code indistinguishable from human creations. Crucially, any product generated by the chatbot will provide a route back to the original FT source material.
This and similar deals may provide one route forward towards a mutually beneficial relationship between content providers and the tech giants.
But where does it leave those small-scale creators and freelancers already struggling to survive? Will they continue to see their product harvested in an enterprise that might ultimately lead to their demise?
The signs are not good. In its scramble for inward investment, including to finance data centres and AI development, the government is under pressure to woo tech titans already playing hard to get.
Google warned last month that the UK risked being left behind in the global race to develop AI unless it set out a comprehensive agenda. That would include creating an environment in which text and data mining was expressly enabled in order to provide confidence to innovators and investors.
