Google's Ad Exchange is demonetizing pages that merely mention issues of race and LGBTQ rights, classifying them as “dangerous or derogatory content”
published by the Atlantic in 8/1963 — after the Google algorithm flagged it as “dangerous or derogatory content” https://slate.com/... @slate : In trying to protect advertisers, Google often ends up grouping news coverage of racism with racist content itself. https://slate.com/... Jonathan L. Fischer / @jon_fischer : Really good piece by @aarontmak on how Google's Ad Exchange can demonetize articles about racism—including MLK's Letter from Birmingham Jail and @Slate coverage of David Duke and, uh, Scrabble https://slate.com/... See also Mediagazer
Context & Ripple Effects
Two months after Google moved against The Federalist and ZeroHedge over race-related policy violations, Slate's reporting shows the same enforcement machinery catching canonical journalism: MLK's Letter from Birmingham Jail and Slate's own coverage of David Duke were classified as "dangerous or derogatory content" by Ad Exchange.
The arc runs back further — in 2017 Google let advertisers target racist and bigoted keywords and even suggested more, so the classifier that demonetizes coverage of racism sits atop a system that once monetized racism itself. That asymmetry is the story: automated brand-safety tools can't tell reporting on bigotry from bigotry, and Google's scale makes its classifier the de facto revenue gatekeeper for news.
First-order effects
- News publishers covering race and LGBTQ rights lose ad revenue on exactly the articles doing that coverage — MLK's letter and Slate's David Duke reporting earn nothing while the words that triggered the flag sit in the text.
- Advertisers buying through Ad Exchange get "brand-safe" placement that is actually sanitized of civil-rights journalism, distorting what inventory the exchange offers.
Second-order effects
- Publishers facing demonetization have an incentive to self-censor — softening or avoiding coverage of racism and LGBTQ issues — or to restructure stories around keywords the classifier tolerates.
- Google's enforcement posture becomes inconsistent in public view: it cracked down on The Federalist and ZeroHedge in June, while researchers had found health misinformation still monetized via AdSense and DoubleClick — pressure to explain why coverage is punished faster than misinformation.
Third-order effects
- When one company's classifier decides which journalism is monetizable, the 2017 warning about Google and Facebook's dominance threatening diversity of ideas hardens from a discovery problem into a funding problem: editorial boundaries get set by brand-safety algorithms.
- The pattern pushes news publishers toward direct reader revenue and alternative ad stacks, since automated moderation at Google's scale structurally penalizes coverage of contested social topics.
The trend: Automated brand-safety classification is quietly becoming the economic editor of news, demonetizing civil-rights coverage while the ad platforms' own enforcement stays asymmetric.