KEY TAKEAWAYS

• AI is a powerful tool, but it is not a strategy. News organisations must integrate AI into their broader goals and values.
• A balanced approach is essential. AI should augment human capabilities, not replace them.
• Ethical considerations and transparency are paramount. News organisations must address copyright issues, protect user data, and ensure that AI is used responsibly.
• New business models are needed. The shift to user-centric distribution and AI-powered search requires new approaches to monetisation and audience engagement.
• Journalists must adapt and upskill. Training and education are crucial for empowering journalists to use AI tools effectively.
• Collaboration is key. News organisations should collaborate, as well as with technology companies and AI specialists, to address the challenges and opportunities of AI.
• Focus on unique value creation and community relationships: In an AI-driven world, being different becomes more important than being better.

AI IS NOT A STRATEGY. AI plays a crucial part in news organisations’ growth strategy. It is vital to embrace it rather than ignore it, but “AI is not a strategy, it’s a means”, Pia Rehnquist echoes Phoebe Connelly, Washington Post, who spoke at the WAN-IFRA Madrid AI Forum, where she referred to AI as the “salt in your menu”, or another “ingredient” in the craft of your news.

AI IS EATING SOFTWARE. AI is becoming the backbone of every process and product, surpassing traditional software in capabilities and adaptability (Petra Ehmann, Ringier)

THE THIRD WAY. Beyond court cases and exclusive licensing with big media brands, there is a “third way” for publishers to find a sustainable economic model for news and LLMS: ProRata, O’Reilly Media, and Miso show the way forward. Consider partnerships or collective action to negotiate better terms with AI companies.

AI AGENTS. Agents and multi-agent frameworks are becoming increasingly powerful. When working together, they can perform complex tasks and dramatically improve accuracy (Petra Ehmann, Ringier). In our newsrooms, AI is not the editor-in-chief. On the contrary, the social networks’ agents guide, determine, produce and promote content without human control in the loop. While AI morphs from being a tool to becoming agent, it creates an ontological difference between social networks and professional newsrooms.

AI AGENTS, take 2. By moving from a simple (interpretative) tool to a generative agent, artificial intelligence raises a major ontological question affecting how information travels on social networks and how users engage with the news and the entire ecosystem. This agent guides, determines, produces and promotes content, making (“editorial”) decisions without humans in the loop. Users are trapped in “cognitive tunnels” locked into cycles of similar content, encouraging engagement but also a form of misinformation (Pierre Louette, Les Échos)

DECLINE OF THE TRADITIONNAL SEARCH REFERRAL. AI-powered search shows 10-52x more usage than a traditional keyword search. Click-through rates to source articles are only 10-15%. “Prompting the prompt” approach drives 80% of daily usage. Click-through rates of 4-10% on suggested questions. Simple chatbots alone don’t drive adoption (Lucky Gunasekara, Miso)

OPEN-SOURCE Media organisations are underrepresented in open-source AI development despite having valuable data and potential use cases. Open-source AI is fundamental to AI development, with many closed models built on open-source foundations. Data quality and structure are crucial for AI model development, often more important than the models themselves. Open-source models rapidly close the performance gap with closed models, especially when fine-tuned for specific tasks. Ethical considerations and copyright issues are essential when using web-scraped data for AI training (Florent Daudens, Hugging Face)

USER-CENTRIC-DISTRIBUTION. The media value chain is shifting, Moving from channel-centric to user-centric distribution models, with power concentrating at the “bookends” – data/infrastructure and user interface/relationships. Formats shift from fixed to “liquid content” that evolves continuously. Prepare for a shift towards adaptive content systems and user-centric distribution where content will adapt dynamically to individual users rather than following fixed channels. Will Platforms become the media, and news media their source.

CUSTOMER-CENTRICITY AND GENUINE COMMUNITY RELATIONSHIP. Engage with AI technology proactively while focusing on unique value creation and creating systems that benefit content creators and consumers (Tim O’Reilly). The way forward? Invest in building genuine community relationships and leveraging human expertise. AI is a powerful tool for enhancing customer-centricity and operational efficiency in news media, and our organisations must maintain a strong commitment to quality journalism (Rien van Beemen).

MONETISATION OPPORTUNITIES. Users strongly prefer direct answers over links, with AI-powered search outperforming traditional keyword search by 10-52 times. A successful model involves offering free “explore” questions while requiring sign-ups for personalised queries, leading to 10-15% monthly growth in subscriptions. AI-generated answers achieve higher CPMs and click-through rates than traditional display ads. Contextual advertising within AI-generated answers achieves higher CPMs and click-through rates than traditional display ads. (Lucky Gunasekara, Miso)

SMALLER IS BIGGER (AND BETTER). Build high-performance models using limited website-specific data rather than training on the entire internet.

HUMAN IN THE LOOP All speakers stressed the need for a balanced approach combining AI with human expertise to enhance journalism and business operations. Future leadership will require enthusiasm, education and the ability to rally teams around these new technologies.

TRANSPARENCY. Scenarios for the future of AI in news span from hyper-regulated AI (‘AI vegan’) to totally autonomous and uncontrollable AI whose outputs are not understood. Transparent AI is needed. While exploiting AI for investigative journalism, AI solutions must be configured directly by users in natural language and based on verifiable sources.

CULTURE. To succeed in AI, care about the culture. Transitioning from a traditional newsroom culture to a tech-savvy environment involves clear communication, team empowerment, and a sound corporate culture.

REDUCE FEAR OF FAILURE: Management should work to alleviate the team’s concerns about making mistakes, emphasising that AI often makes errors but learns from them. Encourage teams to experiment with new ideas without fear of failure. “don’t ask, just do” (Thomas Schultz-Homberg, Kölkner Stadt Anzeiger Medien).

SCENARIOS FOR THE FUTURE OF NEWS. Five potential scenarios INCLUDE: “Machines in the Middle,” where AI gathers information and creates personalised experiences; “Power Flows to Those Who Know Your Needs,” emphasising the importance of understanding audience needs; “Omnipotence for Some, Noise for Others,” highlighting potential extreme bifurcation in information access and understanding; “AI on a Leash,” exploring regulatory and self-imposed constraints on AI capabilities; and “AI with Its Own Agency and Power,” where AI systems become complex and less understood, similar to financial markets. David Caswell, StoryFlow, estimates a transition period from the current state to a new AI-mediated information ecosystem lasting roughly from 2025 to 2035.

LONG-TERM PREPARATION. More than ever, news organisations must improve their organisational flexibility, build direct, authentic relationships with audiences, and focus on fundamental business differentiators (David Caswell, Story Flow).

THE FULL REPORT IS AVAILABLE ON THE WAN-IFRA KNOWLEDGE HUB (For WAN-IFRA members only)