Making AI companies pay people for their for content wouldn’t just be great for the creators – it would boost the industry too, says Dominic Young

You might have seen, on 25 February, that every newspaper front page in the UK was the same: a co-ordinated plea for the UK government to step back from plans they have to exempt AI companies from the usual laws of copyright. The story of how that front page came about has at its heart, of course, money. But it goes deeper than that, to our culture, our creativity and the role that AI will play in everyone’s future.

The current AI hype is being driven largely by investors who think they have spotted the next multi-trillion dollar opportunity: Artificial Intelligence systems which have started popping up all over everyone’s digital life.

The resulting gold rush makes the original California version look like a vicarage raffle. One of the highest profile AI companies, OpenAI, is targeting a $300bn valuation in its next fundraising round. Meta says it is spending $65bn on AI infrastructure this year, Microsoft £80bn.

So AI is cash and resource hungry. It needs four key ingredients. Computer chips, in the form of data centres. Electricity to run them. And smart geeks on high salaries to make it all work. Those three are where most of the investment is directed.

The fourth ingredient in AI is content – work done by other people and published online. It is the fuel which powers the whole thing, arguably the most important and valuable ingredient of all.

Which is why copyright, one of the most venerable, durable and successful legal concepts of all time, has found itself centre stage, again, in the debate about the future of the internet and the fortunes of big tech, society, and our cultural industries.

Hence those front pages, and other stunts like 1000 musicians putting out an album of recordings of empty recording studios. The creators are cross.

Generative AI systems, the kind which power chatbots and image generators, start a bit like a Google search: they scrape the content they find on the internet and store it in giant databases. These are used to “train” the AI.

Unlike search, AI systems generate their own – often highly inaccurate and misleading, but also plausible and convincing – mash-up responses to user prompts, based on (and occasionally slavishly reproducing) the content that was used to “train” or “ground” the AI. They generate no revenue, or traffic, for publishers or creators whose content was scraped, in fact, the opposite: AI users are satisfied by what the AI system gives them and rarely visit the sites which were scraped to generate AI answers.

Unsurprisingly, the people and companies who created all that content, and whose existence depends on humans finding and paying attention to it, aren’t all delighted by this. They would prefer to be asked before their stuff is copied and used in this way, and they would like the option to say no. And if billions of dollars are being invested in these systems which depend on their work, why shouldn’t they get paid?

The best time to have that debate, of course, would be before all the content was copied, processed and used. But that’s not the SIlicon Valley way.

Here’s some advice from Silicon Valley royalty, Eric Schmidt, to AI entrepreneurs:

“What you would do if you’re a Silicon Valley entrepreneur, is if it took off, then you’d hire a whole bunch of lawyers to go clean the mess up, right? If nobody uses your product, it doesn’t matter that you stole all the content. Don’t quote me.”

Following his playbook, the basic training of AI systems has already happened, without anyone being asked or even told before their content was included.

Which is a problem, because AI training is a one-off process. The AI doesn’t “un-learn” if something is removed from its training database. Once content has been used for training, there’s no reverse process other than starting again from scratch. Trying to remove content from an AI would be a bit like trying to un-bake a cake if you wanted to retrieve some of the sugar. So AI companies need everything they have done to date to be legal: anything else would put their whole house of cards at risk of collapse.

Copyright law was created in England in 1710 to deal with a similar issue. The purpose was to stop books being printed and sold without the agreement of the author. Its introduction created the economic incentive which still drives the entire creative sector today. Put simply, it says you can’t copy someone’s work unless you have the right to do so.

You can get the right in a couple of ways. The most obvious is by owning the work being copied, as we all do automatically whenever we create something. Or by being given the right by whoever does own it. Other than a few legal exceptions, none of which cover what commercial AI businesses have been doing, that’s basically it.

If AI loses or concedes that point, though, they have got a big problem. They have already copied and used, reportedly, everything on the entire “open” internet. That’s billions of other people’s copyrights, a big mess for the lawyers to clean up.

If the courts say it’s not legal, they’re going to have to do more than shrug their shoulders and apologise. As well as having to start again from scratch, they’ll be faced with paying huge amounts in compensation. It will probably mark the end of their dreams of a new El Dorado; they’ll be forced to share the spoils with the creators on whose work they depend.

That would actually be a great outcome for everyone, even the AI industry itself – although they haven’t got that yet.

Because, because, because…the introduction of copyright law did not, after all, signal the end of the eighteenth century book business; quite the opposite: with authors able to get their fair share they had very good reasons to write more books. Hence publishers did better, sales went up, public demand was driven sky-high and a society with a few reading dilettantes began to enter the age of mass readership.

An AI sector functioning within the copyright law, acquiring licences and paying the content creators whose work they use, would similarly function much better, and grow much bigger, because it wouldn’t be at constant war with its no-longer reluctant fuel providers – creators and creative businesses.

Yet the AI lawyers and lobbyists would instead prefer to have free access to other people’s work and so have been trying another tactic: get the law changed. They have been busy persuading ministers to make their copyright problem go away. They have found a receptive audience: politicians also have their eye on the AI gold rush, and for a Labour government which has bet everything on economic growth, any prospect of new money is too tempting to ignore.

So the government has proposed that the law should be changed: AI companies should be allowed to keep helping themselves to everything on the internet. The twist, which they believe is balanced and fair, is that copyright owners can opt out – specifically ask AI companies not to copy and use their work.

Nobody needs to ask that right now, because the law does it for them. That means nobody knows how this “rights reservation” will work; the technology will need to be invented – and everyone who cares about their copyright – whether or not they make money from it – will need to start using it.

The creative industries, which will have to carry these new costs with no compensating benefit, are worth £125bn every year to the UK economy. They have huge and proven potential for growth. Undermining their legal foundations and handing their assets to AI companies for free isn’t going to help them achieve it.

Hence the howls of protest, the front pages, the silent albums and 11,000 responses to the government’s consultation on their proposals. Copyright is a form of property, intellectual property, and people don’t like their property being taken from them by law and handed over unconditionally to robbers who want to convert it into profits for their shareholders and massive salaries for themselves.

The particularly idiotic aspect of this is that the goldrush is itself an illusion.

For example, while the big American AI companies like OpenAI, Meta and Google have reportedly spent billions on training their systems, a Chinese startup called Deepseek recently reported achieving similar results for a cost of around $6m. AI training doesn’t look like such a big growth driver after all.

It is hopefully becoming apparent to even the geekiest ministers that using the law to try to tilt the AI game in favour of the UK won’t achieve anything.

Which leaves them with a dilemma deciding what to do.

Allow me to help.

Gazing into my crystal ball, I see a near future in which the best and most valuable content won’t be available for AI without permission. Learning from what has happened, and regardless of what the law says, creators and businesses will use a variety of ways to prevent their work being stolen.

That means that AI companies which want it will need to ask nicely – just as the law already demands. AIs with access to the best content will be better AIs, able to charge more. Creators with the best content will be also able to charge the AIs more, or opt to make their money direct from humans instead.

A market, in other words, which is already developing, will continue to grow.

That might mean a smaller proportion of the money flowing to AI investors because it has been more fairly shared out. That might also result in a more diverse ecosystem of multiple competing AI systems rather than a couple of dominant ones.

But it will certainly mean a bigger market for creators to operate in, new opportunities for the creative industries, an ability for creators to licence and seek revenue from AI companies all over the world, not just in one country, and sustainable growth even as the cost of computing, inevitably, crashes and basic AI technology continues to be commoditised.

Which countries might be well placed to take advantage of that? I can think of one with an immensely successful and disproportionately, globally, valuable creative industry.

Step forward the United Kingdom. We have a chance of real growth, not the smoke and mirrors of the false messiahs of AI. Let’s take it.

Dominic Young is a media strategist who has worked on the interface between content creators, copyright and big tech since the ACAP project in 2006. Currently he is the founder and CEO of the casual payment network for media, Axate.

Subscribers may comment. Please add your idle thoughts using the box below.

Our March/April 2025 issue is available in Waitrose, Waterstones, Smiths Travel outlets, select Smiths high street branches and around 100 indie stores. Click here for your nearest stockist.