Erroneous AI

McClatchy’s AI-driven Content Scaling Agent (CSA) tool, designed to automate content adaptation for different audiences, produced an error in a *Sacramento Bee* article about sexual assault allegations against Congressman Eric Swalwell. The AI-generated version incorrectly merged details from multiple anonymous accusers, despite the original reporting being reviewed by journalists. The mistake was caught and corrected post-publication, though no formal correction was issued, violating McClatchy’s policy. The CSA tool’s flaws extend beyond this incident, with past errors including mixed-up days of the week in evening roundups. McClatchy claims the system enhances efficiency by freeing journalists for original reporting, but critics argue it introduces new layers of work—requiring reporters to verify AI-generated content against their source material. Meanwhile, the *New York Times* faced a separate AI-related blunder when a story about Canadian politics included an incorrectly attributed quote to Conservative leader Pierre Poilievre, later revealed as an AI-generated summary. The error prompted the *Times* to issue a correction and reinforce its policy banning freelancers from submitting AI-assisted content. A *Times* spokesperson denied the guidance was tied to the incident, citing regular policy updates. The cases highlight growing concerns over AI’s role in journalism, where efficiency gains risk undermining accuracy. McClatchy’s Greg Farmer defended the CSA tool, stating journalists reviewed the flawed article before publication, though the error suggests oversight gaps. The incidents reflect broader industry challenges as newsrooms adopt AI tools amid declining resources and rising pressure for digital content.