On 11 March, OpenAI CEO Sam Altman posted on X that he had prompted an AI model to write a metafictional literary story about AI and grief, concluding: “What followed wasn’t bad. I probably couldn’t tell”. The implication was that the AI had attained human competence in the art of literary fiction… “this is the first time i have been really struck by something written by AI; it got the vibe of metafiction so right.”
It’s probably fair to assume that Altman hasn’t spent much time around literary types. If he had, he may have been more careful in describing what most in the publishing industry consider an existential threat so overwhelming, that this year’s London Book Fair in March – the industry’s international annual festival of note – held a number of events dedicated to the topic.
Go deeper with GlobalData
While AI is not sophisticated enough to replace human literary creators just yet, AI companies are working towards this goal, as evidenced by Altman’s post. As AI models become more advanced they require exponential volumes of high quality training data, of which the publishing industry is a treasure trove. And in doing so, these AI companies are engaging in what has been described as ‘the great copyright heist’, a term coined by, CEO of the UK’s Publishing Association, Dan Conway.
During a panel discussion on AI regulation at the London Book Fair, Conway recounted how, when he first began his career in publishing around eight years ago, he had had a ‘weird’ meeting with the UK’s Intellectual Property Office (IPO) about computer generated works provision. “It felt like it was never going to be all that important. And now it’s suddenly massively important,” he says.
It’s not true to say that no one saw AI coming, but on some level the publishing industry, already grappling with low profit margins, a macro shift to ecommerce and a declining reading public, did not. And it is now dealing with the fallout.
Will AI pull off the great copyright heist?
Conway leads an industry organisation representing over 150 UK publishers, whose members are concerned about the UK government’s proposal for a regulatory framework on AI copyright. A dearth of regulation specific to AI has seen technology companies, namely Big Tech, use the intellectual property (IP) of content creators to train their AI models at no cost.
A government consultation was launched in December 2024 and proposals were delivered in Feb 2025. Against this backdrop, Conway has become the de facto ‘face of the resistance’ in the UK against the great copyright heist. Historically, regulation has always lagged behind innovation. While he believes the UK government’s intention to establish a regulatory framework for AI copyright is well intentioned, “the problem has been that the government alighted in the wrong place.”
The government listened to AI companies and creators and essentially came up with a compromise position, something they call ‘rights reservation’, but provision to opt out of having their content hoovered up by large language models is just a conciliator form of IP protection.
“The problem with that, of course, is that there isn’t a technological solution for an opt out and it’s incredibly difficult, if not impossible, to enforce,” says Conway.
“So the worry of the industry, then, is, effectively, you’re creating this complete wholesale reduction in the legal liability for tech companies to zero and saying, access everything, be that, books, journals, film, any kind of content that you want with rights holders, not be able to create that, not be able to police that or stop it,” he adds.
Conway’s position is that ‘cracking copyright open wide’ and offering creators the tech industry’s version of transparency in return is far from a fair deal.
But as sometimes happens in such highly polarising debates, his views on the topic are vastly more measured than headlines suggest. While Conway has been vocal about challenging proposals that endanger the livelihoods of creatives, he also acknowledges that the reality is far more nuanced.
“We’re all using AI. It has incredible potential to make our businesses better, make our people more efficient, get more books into the hand of readers all around the globe,” he says adding that AI isn’t simply wholesale copyright theft.
Is the ‘fair use’ defence really fair?
The industry consensus seems to be that AI will become a friend or foe depending on the outcome of how it is regulated. Indeed, CEO of Hachette Book Group in the US and Hachette UK, David Shelley, told another panel at the London Book Fair that he reserves his optimism or pessimism about AI based on what legislation will be passed in future.
“Our business model only exists on the basis of copyright. Copyright didn’t fall from the heavens, it’s a human construct by lawmakers almost all around the world – and it works,” says Shelley.
Big Tech companies like Meta and OpenAI have pursued the ‘fair use’ defence in copyright infringement legal proceedings claiming that copyrighted original content scrapped and used as training data is transformed to such an extent that it becomes new content that should not require a license. Authors disagree. John Grisham, Jodi Picoult and Jonathan Franzen, among others, have all launched lawsuits against OpenAI.
And on this Shelley is clear: “I think in my 27-year career there has never been a time when public affairs work has been more important, both in the UK and the US. Creative industries are saying to government that there are safeguards that need to be put in place and authors work has been illegally ingested into machine learning.”
If the right regulatory balance is reached, then Shelley says AI could be positive for the industry. “Where AI can help us get more author’s work into the hands of more readers, that’s probably a good thing. But where it competes with authors that’s a bad thing – and there are some grey areas in the middle.”
Broadly speaking, most booksellers will be innately prejudice against AI, even for cover design, says industry veteran James Daunt, founder of legendary independent bookseller Daunt Books, managing director of Waterstones and CEO of Barnes & Noble. “How publishers will incorporate it [AI] will be interesting to see, perhaps at the back end of things like improving customer service for example,” he adds.
Can AI help the publishing industry?
In fact, the ways in which AI tools can assist existing workflows in the industry are endless. Aside from the obvious AI writing and editing tools to draft, proofread and summarise content, AI can help with visual content such as cover design and automated formatting.
AI tools can help get readers with content classification, help publishers with metadata optimisation to get help their titles reach audiences, predict reader preferences and conduct reader profile analysis for better marketing and targeting of reading audiences. And, of course, AI tools for translation is an area that, when refined for accuracy, will transform the translation and distribution of content across borders.
On the administrative side AI will provide the same benefits as it does in any industry in turbocharging mundane tasks, dealing with contracts, market research and more data driven business decision making.
AI can assist in market research, recommend authors or creators, and repurpose existing content for different platforms. This, say techno optimists, will free up time for publishers to focus on the very business of creativity. And for the truly optimistic, AI tools may actually help to enforce IP and protect content creators from copyright infringements.
While the publishing industry is understandably concerned about copyright, the technology industry, overall, continues to move fast and break things with the conviction that technology has the potential to improve every vertical no matter the distribution to its existing business model.
Techno optimism about AI for creatives
At the launch of Google Cloud’s audio generation model Chirp 3’s availability on the company’s Vertex AI cloud platform, Google DeepMind CEO and co-founder Sir Demis Hassabis effused about the application of AI tools for creative industries, saying he was excited about their potential.
“I think new types of tools will be a sort of new way of creating. I feel like the next era is going to be the creative talent embracing these technologies and it will become kind of superpower,” he says.
Hassabis describes this process as a 10x of what creatives were able to do without AI tools. He describes how friends in creative leadership positions, for example, can now create an entire pitch for a new film in a matter of days for a few hundred dollars versus months of work to storyboard something similar. “And I think that allows a lot more creativity to happen. So, I’m very excited about that,” says Hassabis.
When questions were opened to the floor, at the event, the first topic was the thorny issue of IP protections for creatives. On this, Hassabis is positive about the UK’s ability to come to a resolution because of its global stature both in AI development and the policies that are being discussed around managing it.
“This is a very important area that it’s got to be clarified for the field,” says Hassabis acknowledging that Google uses open web data to train its models adding: “We also have opt outs for people as well to take advantage of.”
But the picture becomes far more complicated when considering that most training models draw on a global content pool. “Models get trained in certain places and get deployed all over the place. They’re used everywhere,” notes Hassabis highlighting the need for international standards on the issue.
And it’s here that the UK is in a strong position, according to Hassabis. “We need to be at the forefront of the technology so that we can influence how it goes on the global stage.”
Google’s investment in DeepMind and its leading position in AI development has undoubtedly helped create a whole AI ecosystem in the UK. “We are very strong if you compare us to European countries, for example. And I think that gives us an important voice at the table when this [AI copyright regulation] is going to be discussed.”
The 2024 Bletchley summit established the UK’s leadership role in global AI regulation. The Paris AI Action Summit in February also had regulation top of the agenda. “I think these are really important forums for discussing these kinds of topics. And I’m sure there’ll be an evolution of how this will work over the next few months,” says Hassabis.
As tech companies continue their rapacious pursuit of training data, as regulators scramble, and creatives fight for their lives, the publishing industry approaches its King Canute moment of starting to fully comprehend the futility of railing against the rising tide of AI. Whether the industry can strike the right balance between AI optimism and AI resistance will inform how it mitigates the existential risk to its business while extracting potential value from the emerging technology. However, all may depend on forthcoming AI copyright regulation and, for the moment, uncertainty looms.
