Generative AI in the news business: Not to be trusted

As websites and online news operations try out generative artificial intelligence tools, they’re finding that the output can’t always be trusted.
Gen AI is great for some things such as providing quick definitions for technical terms and acronyms or summarizing information from a trove of government documents. But its habit of providing incorrect information and making things up (“hallucinations”) on some subjects make it not ready to supplant journalists.
One problem with AI chatbots is they present information as though they are certain about it. Chatbots may need to provide a certainty scale with the information they present while they’re in the early days of development.
They also do a poor job of citing they work. Users need citations to find out where AI is getting its information and to fact-check its work.
Already some people and business have committed to generative AI for creating articles and even supposed news websites.
NewsGuard has so far identified 1,271 AI-generated news and information sites operating with little to no human oversight. And it is tracking false narratives being produced by artificial intelligence tools.
“The rollout of generative artificial intelligence tools has been a boon to content farms and misinformation purveyors alike,” NewsGuard said in a report.
Some have described these generative AI sites as producing the equivalent of the meat industry’s “pink slime.”
News media are rushing into generative AI out of a “fear of missing out” (FOMO) on the next big thing. While AI has shown its usefulness in some journalist tasks, it is just a tool. It is simply too early to use AI as an automated front-facing application for news.

Related articles:

The good, the bad, and the completely made-up: Newsrooms on wrestling accurate answers out of AI (Nielsen Lab; Aug. 4, 2025)

Politico’s AI tool spits out made-up slop, union says (Semafor; June 8, 2025)

Business Insider recommended nonexistent books to staff as it leans into AI (Semafor; June 1, 2025)

How an AI-generated guide to summer books that don’t exist found its way into two newspapers (Media Nation by Dan Kennedy; May 20, 2025)

Bloomberg Has a Rocky Start With A.I. Summaries (The New York Times; March 29, 2025)

AI chatbots can’t be trusted, proves study, but Apple made a good choice (9to5Mac; March 11, 2025)

AI Search Has A Citation Problem (Columbia Journalism Review; March 6, 2025)

Photo: Pink slime (Wikipedia)