CNET pauses AI-written articles after errors found in more than half

CNET paused its use of an in-house AI tool to draft finance explainer articles after an internal audit found that 41 of the 77 pieces it had published carried factual errors serious enough to require correction. The tool had been running quietly since November 2022, attributed on-page only to “CNET Money Staff,” until the technology outlet Futurism reported on the practice in January and prompted scrutiny of the outputs.

The errors ranged from minor — incomplete company names, vague or generic phrasing — to substantive miscalculations in articles explaining compound interest and loan terms, the kind of factual mistakes that matter more in personal-finance content than in most other categories. CNET’s editor-in-chief, Connie Guglielmo, wrote that the review was prompted after readers and competitors flagged one article with significant errors, which led the team to check the rest of the AI-assisted batch rather than treat the first case as isolated.

CNET said it would resume using the tool only once it was confident its editorial review process reliably caught both AI and human mistakes, and it committed to clearer disclosure of which articles were AI-assisted going forward. Parent company Red Ventures, which also used similar tools on other properties including Bankrate, faced the same scrutiny.

The episode became one of the most widely cited early cautionary tales about AI in journalism: a case where a mainstream outlet quietly deployed a language model on stories with real financial stakes for readers, without disclosure, and only reversed course under public pressure — a pattern that shaped how other newsrooms subsequently wrote (or delayed writing) their own AI-use policies.