Don't Let Activists or Robots Choose Your Words
From AI chats to the new AP stylebook to a long list of political crazes, online America is constantly looking for ways to put words in your mouth
The 2026 AP stylebook I just received in the mail checks in at nearly 500 pages, ten times as long as the 1951 edition, 200 pages longer than the 1977 guide. The meaning of “style” has expanded massively over the years. Once, the most important part of the book was a “libel guide,” but that’s long gone, replaced by new sections on “inclusive storytelling,” along with enhanced guidance on “race-related coverage,” immigration, AI, and other politically charged topics.
Much of AP style is still good old usage — discreet means prudent, circumspect, while discrete means detached, separate, etc. — but in the 2026 guide, you’ll find a ton of basic descriptive terms shoved into the “avoid” column for questionable reasons, from populist to typical to community to euthansia (but dying with dignity or medical assistance in dying might be okay). Writers are advised not to use a lot of excellent words like “insane” and “demented” and “crazy” outside of quotes, on the grounds of being “derogatory,” while declaring taboo harmless idioms like “turn a blind eye” and “fall on deaf ears.” Meanwhile, a new “inclusive storytelling” section tells editors not to write about diversity and inclusivity — they say that can imply a false “norm” — but to make silent decisions in favor of stories that shine “a brighter light” on the underrepresented.
That’s not necessarily a bad idea, but not long ago no one would have thought of folding that concept into a “style” guide, instead leaving those calls up to editors. Now, the fight to standardize even complex political ideas has become an ingrained feature of American politics, compounded by the explosive rise in AI-generated content. Substack CEO Chris Best, with whom I’m speaking at 5:00 PM ET today — just look for us on Chris’ Substack Notes account — describes hearing complaints from readers about a content landscape not only filled with “soulless slop,” but one where readers are lured to AI-generated prose by unscrupulous humans who don’t disclose that they’re reading something “written by no one.”
I’m looking forward to discussing this issue with Chris, so please come by. It’s a tricky topic with more than a few ethical problems. In the meantime, a few thoughts about the additional layer of deception buried in a lot of this content, as a result of a movement toward jargon-loading in journalistic and official “style” guides:
