AI disclosure – A tech editor details how he uses AI tools in his reporting and writing—plus the boundaries he won’t cross—after AI-related quote errors and corrections raised fresh questions about transparency in journalism.
When a reader comes across a quote that sounds perfectly quotable, it can feel like evidence. But for Benjamin Mullin of The New York Times. the evidence went sideways: in a May 19 report. Mullin said Steven Rosenbaum’s new book. The Future of Truth: How AI Reshapes Reality. contained at least five quotes that appeared to have been manufactured. mangled. or misattributed by AI.
Rosenbaum. who is the executive director of the Sustainable Media Center and a Fast Company contributor. told Mullin he took “full responsibility” for the errors. The episode lands with extra irony given the subject of the book itself. Still, it wasn’t the only public stumble. Earlier in May. The Times issued a correction for an article that had attributed an imaginary quote to a Canadian politician.
That’s the moment the question moves from “what happened to them?” to “could it happen to me?” The editor at the center of this disclosure says he isn’t worried in the specific way the Rosenbaum case is worrying. because he doesn’t insert anything from a chatbot’s response directly into a story draft. If a chatbot offers what looks like a punchy quote—from reporter/podcaster Kara Swisher. for instance—he says he wouldn’t assume it’s real unless he could trace it back to its source.
He’s not arguing that every journalist is tempted to cut corners. But he does describe the lure clearly: if the past has taught writers to look over their shoulders about plagiarism. a “superhuman” writing tool can tempt them to trust output they shouldn’t trust. Rosenbaum’s own reaction, shared with The Atlantic’s Will Oremus, captured that emotion. Rosenbaum said he felt “seduced and betrayed” by AI.
With those failures in view. the editor makes a straightforward case for transparency—showing how AI is being used and. just as importantly. how it isn’t used. His baseline is simple: his wordsmithing is his own, with quality control provided by his Fast Company editorial colleagues. He says thinking about something and writing about it are “pretty much the same act. ” and he “wouldn’t know how to turn part of the job over to an algorithm.” Even when he’s used AI for brainstorming headlines on rare occasions. he says it has felt like more work than it’s worth.
Where AI has a role is in turning research into prose, without replacing the human work of reporting. He admits using Rev to transcribe interviews, Google’s NotebookLM to summarize them, and Grammarly for proofreading assistance. More recently. he says he’s replaced them with similar features built into a bespoke word processor he “vibe-coded” for his own use. He describes the impact on finished writing as modest.
He also uses chatbots every day—especially Gemini—but he frames that use around navigation, not quotation. His goal is not to “get answers from a bot. ” and he says he would not fold chatbot output into his work “under any circumstances.” Instead. he uses bots to point him toward original. human-written sources. often on topics he describes as “arcane.” In his view. the bot is the beginning of the journey. not its destination.
For him. the most consequential AI use is different: he credits Anthropic’s Claude Code as the engine behind software that streamlines the work around his real reporting. Along with his word processor. he says he built a note-taker. an email client. a Bluesky-Mastodon-Threads crossposter. and an RSS reader—apps he says are exactly what he wanted. Every minute those tools save. he adds. goes back into interviews. other research. listening to what people tell him. and writing.
He isn’t presenting his approach as the only one that can work. He points to Alex Heath. founder of the Sources newsletter. who told Maxwell Zeff of Wired that he “snag[s] scoops” and leans heavily on AI for the writing aspect of his work. The editor says he appreciates that disclosure and is impressed by the results—along with the hope that more people in the business will learn to use AI in ways that never leave them feeling “sheepish. or worse.”.
The newsletter closes with his own editorial identity: you’ve been reading Plugged In. Fast Company’s weekly tech newsletter. from global technology editor Harry McCracken. He invites feedback by email at hmccracken@fastcompany.com. and he notes he is on Bluesky. Mastodon. and Threads. with Plugged In also available on Flipboard.
AI in journalism disclosure hallucinated quotes media corrections Steven Rosenbaum Benjamin Mullin The New York Times correction Sustainable Media Center Gemini Claude Code Rev NotebookLM Grammarly newsroom transparency