Wired: Newsrooms Are Quietly Rolling Out AI Writing Assistants — and Staying Silent on the Consequences

AI writing is quietly seeping into newsrooms, not through loud announcements but through small concessions: first a draft of a news brief, then a summary of financial data, then the first version of a field report. Publishers call it efficiency. Journalists see something else.

Wired raises a subject the media industry prefers to sidestep: what happens to journalism when AI takes the first, and most important, step in producing a story? The draft sets the structure. The draft determines the angle. The draft is already half of a finished article. If a machine writes it, who is responsible for what ends up on the page?

Pressure on newsrooms is mounting from several directions at once. Ad revenue keeps falling. Readers increasingly get their information from social networks and search engines, often bypassing the publication's site entirely. Against this backdrop, cutting the cost of text looks attractive: AI tools promise the same speed with lower staffing costs.

BuzzFeed, CNET, Sports Illustrated and G/O Media have all experimented with automated text in one way or another over the past two to three years. The results are mixed: there were factual errors, outright failures, and quiet retreats from experiments without any official explanation.

But the problem runs deeper than factual errors in AI drafts. Journalism does not exist because someone needs to assemble facts quickly into a single paragraph. It exists because a living person with experience, sources and a point of view can find a story where no one is looking, ask the uncomfortable question and take responsibility for every word. AI can imitate the structure of a text. It cannot imitate judgment.

Another hidden cost is reputational. When it emerges that a piece was written or assembled by a machine, readers' trust in the outlet drops, and it recovers extremely slowly. CNET learned this in 2023, when it came out that dozens of financial articles had been produced by AI without proper disclosure. The newsroom had to pull some of the publications and introduce a mandatory label. The lesson was painful and expensive.

Tech companies selling tools to newsrooms prefer a different vocabulary. They talk about "accelerating the workflow," "reducing cognitive load," and "assistance, not replacement." In practice, the line between "assistance" and "replacement" is blurry and keeps shifting. Today AI writes a draft of a news feed item. Tomorrow, an analytical review. The day after, an opinion column with a living author's name on the byline.

The reaction of journalists themselves is telling. In newsrooms where AI tools are introduced without open discussion with the team, quiet resistance grows. People feel their professional identity eroding. When your job is to edit someone else's draft rather than create your own, you stop being a journalist in the full sense of the word.

What is happening is not just a question of technology or economics. It is a question of what journalism is as an institution. If newsrooms accept AI writing as the norm under pressure from short-term financial incentives, they risk destroying exactly what sets them apart from algorithmic news aggregators. Readers choose living publications not for speed, since bots won that race long ago. They choose them for trust, for voice, for the judgment of a specific person. That cannot be automated, and attempts to do it covertly sooner or later become obvious.