The JournalismAI Festival 2025 wasn’t a mere celebration of cool new tools: it was an open, and often painful, engagement with the industry’s existential crisis. The message was in unanimous one from journalists, editors and technologists: the role of news media in AI has utterly transformed—news is increasingly no longer the output as many have presumed; instead news stories are now busy becoming input—or rather, just ingredients—for these Large Language Models (LLMs), which run search and generative AI systems.
‘The bigger play that’s happening is that media is getting added to AI, and we are all becoming part of AI systems,’ one industry insider put it bluntly. This transition has deep implications for an industry that was already staggering from two decades of digital disruption. The issue is one of visibility and value extraction.
The Great Unbundling: Traffic vs. Content
For a long time, the “old deal” made between publishers and platforms was purely transactional: Publishers got traffic (clicks), in return for their content. The development of generative AI narrative models and of AI Overviews in search has fragmented this dynamic.
Zero-Click Answers: Search engines reliant on LLMs increasingly offer answers to search queries directly on their search page in a clean, definitive format (AI Overviews). These responses are often made up of previously reported news summaries and information without having to touch the publisher’s original website, which makes them a boon.
Traffic Collapse:Big publishers in countries such as Brazil, South Africa and Indonesia are experiencing a traffic collapse of between 50% to 60% year-on-year due to the changes. Without traffic, the critical funnel for attracting subscriptions and advertising revenue breaks down. “Search and social gave us traffic; A.I. doesn’t,” said one media executive.
Read also:
- ChatGPT 5.1 has a New competitor
- AI could replace 3 million jobs
- Gen-Z innovators turn down Musk’s multimillion dollar offer
- PhonePe Announces Strategic Collaboration with OpenAI
The Licensing Dilemma
In a bid to reclaim some of that value there are news organisations who have signed content licensing agreements with AI developers. These flat fees provide attractive and immediate revenue, experts cautioned, but it could camouflage an existential threat. In their efforts to prevent bad actors from finding audiences on their platforms, publishers will be contributing the entirety of their content archives to train the tools that eventually keep people off of their platforms, leading them potentially toward a disincentivization to create quality, investigative journalism down the road.
The New Toolkit: Add — Not Replace
Newsrooms are not standing still, however, despite profound concern about the business. The Festival demonstrated amazing innovation that all centered on augmenting human journalists, not getting rid of them.
Noticing the Needle in the Haystack
AI is being intentionally deployed on jobs impossible for humans to do at scale:
Investigative Journalism: Tools are being developed to sift through vast quantities of information (such as political party manifestos, as shown by one UK fact-checking project) and uncover story leads and track government accountability.
Enhancing Audience and Efficiency
Engagement boosters: One UK publisher used a language learning model (LLM) to create prompts for discussion around their articles – and saw commenting from readers rise by 3.5% – revealing the that with some thought, AI can be deployed to enhance those direct reader relationships everyone is clamouring for.
Workflow Automation: Everything from automated transcription and first-draft translations to generating “rough cuts” of video, audio or even photo stories, AI-efficiency frees up human journalists to think critically-judiciously-ethically.
The Path Forward: Trust and Transparency
The consensus was trust and all agreed, in an AI-mediated world the news industry only has one real point of differentiation and that is trusted content.
Publishers can’t even really win when it comes to explaining how and when AI is used; of course they should be transparent, but there are nuances — some subscribers have said that reminders about AI in articles make the journalism seem “cheap” by comparison. The solution lies in AI literacy, newsroom-wide, combined with a rededication to the basic tenets of journalism — accuracy, fairness and human oversight.
It is no longer a battle to just acquire eyeballs, but trust — the one scarce resource that generative AI, with risk of “hallucination” and bias baked into its models, as it still cannot reliably produce.
