A new column from The Verge argues that AI writing detectors are ushering in what it calls "a new era of distrust" — a moment when any piece of text, from a student essay to a work email, can be flagged as machine-made and its author left to prove otherwise.
The piece runs in The Stepback, The Verge's weekly newsletter that breaks down a single essential story from the tech world. It is credited to Emma Roth, and the newsletter reaches subscribers' inboxes at 8AM ET. The column opens by reaching back in time, noting that the suspicion at issue predates ChatGPT's rise rather than beginning with it.
Beyond that framing, the available excerpt is brief, and it does not include the specific detector tools, accuracy figures, or institutional policies the full column goes on to discuss. What is clear is the argument's shape: the problem is not only whether AI can write convincingly, but what happens to trust once software claims to be able to tell.
That distinction is the reason this story lands beyond tech circles. Detection tools are already woven into classrooms, hiring pipelines, and publishing workflows, and an accusation from one carries real consequences — a failed assignment, a rejected application, a damaged reputation — for a person who may have no practical way to demonstrate their own authorship.
Why it matters: when the burden of proof shifts onto writers to show a human wrote their words, the cost of a wrong flag is paid by people, not by the software that flagged them.