AI makes reliable journalism harder

On March 17, the Reuters Institute for the Study of Journalism gathered reporters, editors and researchers for its AI and the Future of News conference in Oxford, where speakers said generative AI was reshaping reporting, fact-checking and the way audiences encounter news. More than 3,000 people signed up to follow the event, according to the institute’s summary of the conference. The discussion landed as editors, fact-checkers and publishers were already dealing with a surge of synthetic text, altered images and AI-generated video across major platforms. April 3 brought a sharper formulation from Poynter’s International Fact-Checking Network, whose authors wrote that AI-generated content, chatbot errors and shrinking newsrooms were making reliable information harder to find. The problem they described was not limited to one fake clip or one fabricated article. It was the accumulation of synthetic material, mixed with authentic reporting, that made verification slower and trust harder to sustain. (reutersinstitute.politics.ox.ac.uk) ### Why are editors spending more time checking material that once looked routine? March 18 reporting from the World Editors Forum’s Editors Weblog said manipulated footage that once required “Hollywood budgets” could now be produced on consumer hardware, raising the stakes for routine verification work. The article said larger organizations could assign dedicated teams and tools to suspicious material, while smaller newsrooms often lacked the same capacity. That gap matters because verification now extends beyond obvious hoaxes to ordinary-looking clips, screenshots and eyewitness posts. (poynter.org) The Associated Press has responded by formalizing newsroom rules and building verification tools around them. AP says accuracy, fairness and speed remain its guiding values and that AI will not replace the central role of journalists in gathering and ordering facts. In late 2025, AP introduced AP Verify, a browser-based platform designed to help authenticate photos, videos and text in newsroom workflows, according to reports on the launch. (editorsweblog.org) ### What is making the information environment harder to navigate? April 3 commentary from Poynter identified three pressures at once: actors using AI to disinform for political or financial reasons, people relying on chatbots for news despite hallucination risks, and a broader erosion of what counts as evidence online. The authors wrote that even a system that is wrong 5% of the time is hard to trust if users cannot tell which 5% is false. Their argument was that AI does not just create more false material; it also makes uncertainty itself more common. (ap.org) January 9 reporting from NBC News described that uncertainty during fast-moving events, when AI-generated images, old footage and altered visuals circulated alongside real material. Stanford Social Media Lab founder Jeff Hancock told NBC that AI could undermine the “trust default” that leads people to believe communication unless they have reason not to. He said the immediate challenge would be that people may stop trusting what they see in digital spaces. (poynter.org) ### How big is the volume problem for publishers and fact-checkers? NewsGuard said in March that it had tracked 3,006 AI-enabled content farm sites publishing unreliable information and misinformation. The company’s tracker said those sites operate across multiple languages and are part of a broader effort to use generative AI to scale false or low-quality content. Other coverage citing NewsGuard’s work said hundreds of new sites were appearing each month. (nbcnews.com) The Center for News, Technology & Innovation said on March 19 that generative AI offered productivity gains for newsrooms but also posed risks of inaccuracies, ethical problems and damage to public trust. CNTI said the policy response remained unsettled, leaving publishers and technology companies to set many of the practical rules themselves. That has pushed newsroom leaders toward internal guidelines, public disclosures and training rather than waiting for a settled regulatory framework. (newsguardtech.com) ### What are news organizations doing in response? BBC Verify has turned verification itself into a visible editorial product. A March 5 BBC Verify explainer said its journalists use four methods to identify AI-generated images and videos, while BBC reporting on related initiatives has emphasized showing audiences how verification is done. That approach reflects a broader newsroom calculation: when synthetic media becomes easier to produce, publishers may need to make the reporting process more legible as well as more rigorous. (cnti.org) February 9 commentary from Digital Content Next said newsroom leaders were emphasizing clear AI policies, staff training and editorial judgment as generative systems moved deeper into workflows. The Reuters Institute conference on March 17 put the same issue in public view, with journalists from Bloomberg, the Bureau of Investigative Journalism and the Pulitzer Center discussing how AI is covered and how it changes fact-checking and reporting. (youtube.com) March 17 is the clearest next marker in this story because the Reuters Institute has already framed AI, fact-checking and newsroom practice as a continuing reporting agenda, and organizations including AP, BBC Verify, Poynter and NewsGuard are publishing new verification guidance and tracking updates as the year advances. (reutersinstitute.politics.ox.ac.uk) (digitalcontentnext.org)