If AI uses journalism, AI must pay: protecting the news that underpins democracy

Publishers around the world are sounding the alarm: AI chatbots are using professional reporting to answer users’ questions while sending back less traffic and less revenue. Deutsche Welle’s reporting captures a question that matters beyond newsrooms — should the companies that build these AI tools pay for the journalism they rely on? The short answer is yes. The how is the harder but urgent part. Quality journalism is costly. Investigative reporting, local bureaus, fact-checking and legal safeguards require time and money. That work underpins public accountability, exposes corruption, informs voters and helps communities make decisions. When AI systems scrape news content to train models or answer queries without returning traffic or revenue to the original publishers, that supply chain is broken. The result is an economic squeeze on outlets already struggling in a digital world. Publishers argue — correctly — that they are being used as a kind of unpaid research department. Their reporting feeds the models that power assistants, yet users often get summaries and answers without seeing links, citations or paywalls that would bring readers back and fund future reporting. If that dynamic continues, outlets will be forced to cut staff and close bureaus, eroding the public-interest journalism that democracies depend on. Tech companies counter that AI offers value in different forms: summarisation, accessibility and new products. There is truth in that. But value does not absolve responsibility. When a company’s product depends on distinct creative labour, it is reasonable to expect compensation or licensing arrangements. The history of media rights — from broadcasters paying for sporting feed rights to platforms licensing music — shows that markets can adapt when there is clarity and political pressure. Governments and regulators have a role to play. They can require transparency about training data and compel citation standards that make it obvious when AI answers derive from specific news reports. They can also consider frameworks that require contribution to a journalism sustainability fund or mandate licensing negotiations — models that have precedent in other sectors. Any such policy must balance innovation with protection for the public interest. Publishers, too, can act strategically. Aggregation-friendly metadata, clearer licensing offers, APIs that provide paid access for developers, and experimenting with products that integrate AI while preserving subscription pathways are practical steps. But such commercial measures will work only if regulators create a floor under which public-interest journalism cannot fall. The stakes are not abstract. A world where AI can answer all questions but the organisations that produce independent reporting have been hollowed out is a poorer information environment. It is easier to generate facts; it is harder to verify them. If we value accurate, probing and locally relevant news — coverage that explains council decisions, holds hospitals and schools to account, and uncovers wrongdoing — we must ensure the economics support it. The debate should not be framed as tech versus media. It should be about sustaining a healthy information ecosystem in which AI tools can offer convenience without undermining the supply of quality reporting they rely upon. That will mean new business models, better transparency and, crucially, policy interventions that recognise journalism as essential infrastructure for democracy. If AI is using journalism to stand on its shoulders, then it should be willing to pay for the lift.