On July 20, YouTube clarified the rules governing which videos may carry advertising, and among the three flavors of “inauthentic content” now barred from monetization is a category the company calls off-putting — material engineered to distress or manipulate viewers. The example TechCrunch pulled out of the policy language was precise: a video featuring an animal in distress, followed by someone who arrives to rescue it. This week, the top link on Hacker News — 666 points, 209 comments — is a video about a woman who rescued an injured crow and now receives small thank-you gifts from the local birds. A crow leaving trinkets for the human who patched its wing is either the most charming thing on the internet or an exact match for the pattern YouTube just demonetized. The platform cannot tell the difference, and it has stopped pretending to try.
The Word Doing All the Work Is ‘Off-Putting’
Note who is supposed to be offended. A YouTube spokesperson told TechCrunch the company had heard from viewers that they find this content distasteful. Maybe a few did. But viewers, in aggregate, are the people who made animal-rescue video one of the most-watched genres on the platform, and who this week pushed a crow with a gift habit to the front page of a site that is not sentimental about anything. Nobody is put off. The parties reliably put off by a distressed animal are media buyers, brand-safety vendors, and the people who sell them spreadsheets.
The tell is that the three banned categories are not three kinds of bad video. Two of them are fraud categories — synthetic media passed off as real, mass-produced output from a content farm. The third is a feelings category. And the enforcement record confirms which one the machinery was built for. In December 2025, YouTube terminated two large channels, Screen Culture in India and KH Studio in the United States, for repeatedly posting AI-generated fake movie trailers. Both had subscriber counts in the millions. That is deception, and nobody should mourn it. The new rule reaches content that is neither fake nor deceptive, merely upsetting — which is a different crime, prosecuted in a different court, with a different burden of proof. Specifically, none.
The Right Wants Two Things It Can’t Have
If you believe, as much of the right now does, that platforms are common carriers — that they must carry what they carry and stop editorializing — then you have to accept the corollary that goes unspoken at the panel discussions. A common carrier cannot be compelled to sell advertising against everything it carries. Carriage and monetization are separate businesses with separate customers. The audience is not YouTube’s customer. The advertiser is. Demanding that the platform stop discriminating among videos while also demanding that it stop letting advertisers discriminate among videos is asking for the state to set ad rates, which is a thing conservatives opposed until roughly 2016.
A second-year brand-safety analyst at a mid-size agency, reached on the phone during a break at an industry conference in Manhattan last month, put the mechanics more bluntly than any think-tanker has: “Nobody in my building has ever watched a rescue video. They get a spreadsheet with a flag on it. The flag says animal distress. It doesn’t say fiction.” That is the entire system. It is not a judgment about art or truth or even slop. It is a keyword match against a client’s exclusion list, and the client’s exclusion list is written by people who have never met a crow.
The Rule Can’t Tell a Rehab From a Racket
YouTube now evaluates whole channels rather than individual videos, which is efficient if your goal is to catch operations running the same template forty times a week. It is catastrophic if your goal is precision. A wildlife rehabilitation center posting an honest intake video — a hawk with a shattered wing, an owl that lost an argument with a car — teaches the classifier the same lesson as a staged duck rescue in a drainage ditch outside Bucharest. The rehabilitation center has three employees and no compliance consultant. The slop farm had twelve channels and has already pivoted to something else.
The collateral is the woman with the crow. She did not stage anything. She did not ask the birds for gifts; the birds brought them. Under the July language, her account is one emotional-manipulation flag away from a demonetization notice, and the notice will not explain which video did it, because the system is designed not to.
Say the Quiet Part Loud
There is an easy fix, and its near-total absence from the discussion is the strongest evidence that the discussion is theater. YouTube should stop dressing a commercial decision as a taste decision. Say advertisers won’t pay for this, instead of viewers don’t like it. Then creators know the rules, advertisers get their exclusions, and nobody has to pretend a spreadsheet is a referendum.
The current formulation asks audiences to believe their own preferences are being protected from themselves. It asks creators to guess. And it asks us to accept that the woman with the crow is a policy problem rather than the best thing on the internet this week. The crow, for what it’s worth, keeps paying its invoices on time. It never checks the numbers.
Sources
- Development of cognition in corvids | bioRxiv
- Crows Can Count Aloud Much Like Toddlers, New Study Finds | Audubon
- Cognitive Showdown: Crows vs. Toddlers in the Science of Intelligence – ERIC DORFMAN
- YouTube’s AI content crackdown in 2026: What changed, who is at risk, how to adapt – ScaleLab
- YouTube AI Crackdown 2026: What Creators Need to Know | OutlierKit
- YouTube clarifies policies around AI slop and upsetting videos | TechCrunch