One of the thorniest AI issues for publishers is how to respond to tech companies scraping content to train their LLMs. Should they sue to prevent their copyrighted work being used without permission or payment, or should they license their IP to the tech giants in the hope that sustainable commercial models develop?
At our Definitive AI Forum, held in London at the end of 2025, we asked a group of experts to share their insights into how publishers should be responding to AI models using copyrighted content, and discuss what strategic options exist to protect their interests.
Moderated by media consultant Paul Hood, the panel featured Tami Hoffman, Director of Public Policy at The Guardian; Amir Malik, AI & Digital transformation lead at consultants Alvarez & Marsal; and Sajeeda Merali, CEO at the UK publisher’s association, the PPA.
Hood laid out four topics for discussion: how to confront the problem of LLM’s taking content without permission or payment; whether bi-lateral licensing or content marketplaces offer a solution; the role of regulation in protecting copyright; and the impact of agentic AI and bot traffic on publishers who rely on human audiences.
You can listen to the full panel below, or by searching ‘The Publisher Podcast’ on your podcast app.
If there was one clear takeaway from the panel, it was that there is no single solution to the challenge posed by the training of LLMs on copyrighted content. That doesn’t mean, however, that publishers are powerless in the face of AI. A mix of collective action, better regulation and a cautious acceptance of commercial realities will give publishers a way forward.
Stronger together
The panel agreed that a unified approach from the industry was the best way to make sure that publisher concerns around AI and copyright were heard. “I think that having a collective voice is important,” said the The Guardian’s Hoffman. “If you look at something like the Make It Fair initiative where all publishers gave over their front pages… that was something that was open to all publishers, large and small.”
She said now is the time to be ‘good, active’ members of the publishing industry’s representative bodies. “I think sitting back and hoping that someone else is going to pick up the mantle and run with this – none of us can afford to do that.”
The PPA’s Merali underlined this, saying how important it is for media organisations to engage with the issues. She advised senior leaders, as generalists with an overview of their operations, to lean in to understanding how LLMs are impacting their businesses. “Capturing that data, sharing it with us is super important. It really elevates how we’re able to lobby on publishers behalf,” she said.
Malik agreed, saying that collective action will be more effective than a case-by-case approach. “We’ve got to move quickly into a more collective model, and we’ve got to use spheres of influence to break down the lobby of the platforms, to actually make them comply with publishers. If you don’t do that, it’s very difficult to see how you get out of this.”
Tami gave publishers credit for doing ‘a fairly good job’ of putting some of their traditional rivalries behind them to try and approach the UK government in a collective way. “It’s not perfect, but I think there has actually been some quite good collective action,” she said.
Legal actions & licensing
One area where publishers have decisively taken collective action is in the class action lawsuits they have brought against a variety of AI platforms.
The Guardian is involved in a court case against the AI startup Cohere in the United States. “We’re one of 14 publishers involved in that lawsuit,” Hoffman explained. “There are 53 court cases at the moment going through the courts, and that’s where copyright is going to have to be tried. Although it’s an expensive and time consuming route, publishers do have to stand up and be counted before it’s too late.”
Licensing deals are the flip side of legal action, with individual publishers doing deals to give LLM developers access to their content.
Saj also reminded the audience that there are already services that exist to collect money on behalf of publishers, citing the PLS (UK Publishers Licensing Service) and the CLA (UK Content Licensing Association). “It is just about trying to create a model that is fit for purpose around training, but the pipe work exists.”
She acknowledged that the LLM licensing landscape is complicated with publishers doing their own bi-lateral deals at different levels and with different outcomes. “That doesn’t stop us coming together as an industry on collective licensing,” she said.
Hoffman agreed, saying that publisher groups have a role to play in setting industry norms and standards for how publishers want their content to be licensed. That’s why we need to be around the table to make sure that the norms, the terms on which these things get set up, have a strong publisher voice.”
Hoffman can see a future where multiple content marketplaces thrive. However, where there is no issue with the supply of content, there is a problem with demand. “There isn’t the right incentive scheme at the moment for most LLMs to come to the market. Why pay for content if you can take it for free?” she explained.
Part of the solution could be technical blocks on LLMs: “I don’t think [they’re] ever going to solve the problem, but they can create friction. They can start to cut that supply, which will hopefully bring people to the table.”
Regulation is another part of the solution. “I also think that there is going to be a need for some kind of statutory, mandatory backing,” said Tami. “Even though we have new standards coming in, if they’re not legally enforceable, they’re going to have a limited impact.”
However, she thinks that until regulators tackle Google’s monopoly position in search, the question of AI legislation can’t be resolved. She explained that Google uses the same bot to index for search and to create its AI summaries: “As a publisher, you can’t switch it off for one of those functions without switching off the other. It’s totally unviable to switch off Google bot, because then none of us would surface in search.”
“What’s the point in having legislation about bot blocking if publishers [can’t] block Google? Where this conversation goes is predicated on what happens to Google and whether any regulator is bold enough, is politically empowered enough, to tackle it.”
The final piece of the puzzle for Hoffman is for publishers to engage the platforms commercially and points to the area of enterprise licensing as a way in. She explained that the absence of rights cleared enterprise AI was creating a liability gap in the market and that it could be attractive to LLMs to license content so that they have legal surety.
“I think that’s when you start to see the dial really shifting,” she explained, “When it stops becoming them and us, when there is mutual benefit to actually sitting around a table and creating a solution that works in varying degrees for both sides.”
Listen to the full session above, or by searching ‘The Publisher Podcast’ on your podcast app of choice.
Header image generated via ChatGPT
