At a major metropolitan newspaper, a senior editor assigned a routine rewrite of a city council press release last month—but handed the task not to a junior reporter, but to an artificial intelligence tool. The AI generated a clean, factual summary in 12 seconds. The editor made three minor corrections. The story published with no byline and, for now, no human complaint. This moment, replicated in newsrooms worldwide, marks a quiet but profound shift in how journalism is produced, raising urgent questions about accuracy, accountability, and the future of newsroom jobs.
The technology behind this transformation is generative AI, specifically large language models trained on vast datasets of published news, books, and web content. These models can summarize, rewrite, and sometimes generate original articles from raw data like earnings reports, sports scores, and government records. According to a 2024 Reuters Institute survey, over 40% of news executives reported using AI tools for some aspect of content production, a sharp rise from just 10% two years prior.
News organizations are deploying AI in two main ways. The first is straightforward automation: turning structured data—such as a school district’s test results or a list of property sales—into short, formulaic articles. The Associated Press has used this approach for quarterly corporate earnings since 2014, producing thousands of stories annually that would have required weeks of human labor. The second, more controversial application is generation: producing longer, narrative-driven pieces on topics like crime trends, local election recaps, or health advisories, with minimal human oversight.
“AI is not a writer; it’s an expensive autocorrect,” cautioned Dr. Elena Marchetti, a media ethics researcher at Columbia University. “The danger is when editors trust the output too quickly. These models still hallucinate facts, misattribute quotes, and struggle with nuance. A rewrite of a police press release could amplify a biased portrayal if no human checks the original source.”
Data backs Marchetti’s warning. A 2023 study by the Tech Policy Lab at the University of Washington found that AI-generated news summaries contained uncorrected factual errors in 23% of cases, with errors often subtle enough to slip past tired copy editors. “It’s not terrible, but it’s not good enough for a professional product,” one participating editor told researchers.
For readers, the implications are immediate. When an article is AI-generated—even if rewritten from a human-sourced original—verifiability becomes opaque. Where did the facts come from? Was the original story itself accurate? Most outlets currently do not label AI-assisted content, citing fears that “machine-made” tags will erode trust in an industry already struggling with credibility. Consumer advocacy groups, including the Society of Professional Journalists, now call for mandatory disclosure labels on any article substantially produced with AI.
The broader impact extends beyond editorial rooms. Newsrooms that adopt AI for rewriting risk eroding the very skill—writing and fact-checking—that builds a reporter’s craft. “You cannot outsource the learning curve,” said Fiona Torres, a former city hall reporter now training AI models for a tech firm. “If you never write a first draft, you never learn why it matters to call the councilmember again for clarification.”
Next steps are taking shape. Several major outlets, including Reuters and The Wall Street Journal, have implemented strict policies limiting AI to basic data aggregation and requiring human approval for any content that cites named individuals or original reporting. Industry bodies are also developing best-practice guidelines, with a draft from the World Editors Forum expected later this year.
For now, the technology is a tool, not a replacement. But the onus falls on editors to wield it wisely—and on readers to stay vigilant. As one veteran editor put it: “If you can’t tell whether a human or a machine wrote the story, that’s not a sign of success. It’s a sign you’re not paying attention.”
