AI is ‘existential’ for journalism, says WAN-IFRA expert

Artificial intelligence is no longer optional for newsrooms – it’s existential. That was the message from Ezra Eeman, WAN-IFRA’s AI expert and director of strategy and innovation at Dutch public broadcaster NPO, during a recent industry webinar focused on the future of journalism.

Eeman led a wide-ranging discussion on how publishers are responding to the rapid evolution of AI and the strategic shifts required to stay relevant. He shared insights from a January study tour to San Francisco, where media executives met with OpenAI and other startups to assess the state of the industry in 2025.

One of the key takeaways was that the traditional sources of traffic – search and social – are in decline. Newsrooms, Eeman said, must reposition themselves within emerging AI ecosystems to ensure visibility. Yet monetisation remains elusive. “Partnerships are scarce,” he said, noting that OpenAI has a small global partnership team – reflecting the limited maturity of the commercial AI market.

Eeman also pointed to the concentration of power within big tech. The top seven companies involved in AI now account for about 30 percent of the US economy, prompting concern from some investors about a potential AI bubble. At the same time, the innovation pipeline remains fragile and unevenly distributed.

Open-source communities and smaller newsrooms may offer alternative paths. Nikita Roy, founder of Newsroom Robots Lab, said the journalism sector lags well behind others in adopting the latest AI tools. “Tools like Langchain are being used across industries to build custom AI applications – but hardly anyone in news is using them,” she said. The problem, she added, is uncertainty about return on investment. Many organisations remain stuck in isolated pilot projects without long-term strategies.

Florent Daudens from Hugging Face agreed, urging publishers to move away from treating AI as a bolt-on or productivity shortcut. “Newsrooms need to understand AI systemically – not just as a tool for individual tasks,” he said.

One case study came from local Norwegian publisher iTromsø, which developed an AI assistant named DJINN to process and summarise municipal documents. The tool – designed and implemented by editorial staff – sped up reporting and improved coverage, an example, Roy said, of how editorially driven innovation can work.

The webinar also addressed changing patterns in news consumption. Eeman warned that AI agents capable of summarising stories without linking back to publishers could upend business models built on traffic and advertising. “If users get the information without visiting your site – the economic model collapses,” he said.

Daudens raised concerns about the accuracy of AI-generated content, citing a recent BBC study that found many outputs lacked basic factual reliability. High-quality training data, he said, is essential – and this is where publishers can play a vital role.

The legal landscape is also shifting. Roy pointed to the Reuters v Ross Intelligence case as one that could set a precedent for how copyrighted journalism is used in AI training. But legal action alone won’t be enough. “Publishers need to engage with the AI ecosystem through partnerships and open-source collaboration – if they want a seat at the table,” said Daudens.

The discussion made clear that the industry is at a crossroads. Newsrooms must move beyond experimentation and invest in long-term AI strategies that support both editorial integrity and business sustainability.

Source: Noah Wire Services