In January 2023, it emerged that CNET had been publishing personal-finance explainers written substantially by an AI tool, with a disclosure note most readers would have had to go looking for. When outside reporters checked the work, CNET ended up issuing corrections on more than half of the roughly seventy-seven articles reviewed — errors in basic financial facts that a competent human editor should have caught before publication, disclosure label or not.
That case is instructive precisely because CNET did disclose. The failure wasn't secrecy; it was that the label answered one question — a tool was involved — while leaving the harder ones unaddressed: what got verified independently of the tool, and what editorial process, if any, sat above the output before it published.
Sports Illustrated's episode later that year was arguably worse, because there wasn't even honest disclosure: reporting found the outlet had run product-review content under fabricated author names with AI-generated headshots, through a content partner, with no indication to readers that the bylines weren't real people. A reader comparing the two cases learns the same lesson from opposite directions — a label without a standard and a byline without honesty both leave the reader with less than they think they have.
The absence of a shared standard means every newsroom quietly defines "AI-assisted" differently, which makes the label nearly useless for comparing across publications. A reader cannot use the same words from two different outlets to infer the same thing about how carefully a piece was checked.
The Associated Press's published generative AI standards are closer to what the industry actually needs: a document that spends more space on what must be verified and what AI tools are not permitted to touch — quotes, sourcing, judgment calls — than on writing style. That separation, verification standard distinct from disclosure label, is the part CNET's episode skipped.
At Duck Times, that separation is the actual commitment: naming the writing source on every piece, and treating facts, quotes, and figures as either checked against a primary source or explicitly flagged as unverified in the piece itself. Here is the falsifiable version of that claim: if a Duck Times piece is ever found to contain a fabricated statistic, quote, or source presented as verified, that is a correction owed to readers, not a disclosure-label technicality to hide behind — and readers are entitled to hold us to exactly that standard, not a softer one.