Sam Oh, the Vice President of Marketing at Ahrefs, recently shed light on this capability. Oh disclosed that a video posted by Ahrefs was flagged by YouTube for a rather unexpected reason. The video in question displayed a snippet from a book, and within that snippet was the name “Donald Trump.”.
Following this, YouTube flagged the video under its “Election advertising in the United States” clause. Ahrefs was subsequently prompted to “Review and fix ads that violate ads policy to update the status of your campaign.”
old news. the transcript commonly mistakes words in videos for swears or racial slurs and gets people demonetized
It’s not a secret; they’ve been doing this for a long time. It’s why you see a lot of YouTubers who cover news stories black out certain words if they share snippets of a news article.
What’s weird about it is that the policy doesn’t seem to be the same for text as it is for audio. For instance, Philip DeFranco has to censor out words like “rape” from articles he puts up graphics of, but doesn’t have to censor it out as he narrates the article verbally.
Is anyone surprised?
We knew they could do automatic captions (Don’t know why don’t offer that standard any more). We know they monitor what’s in the video, meaning the words. We know this was an automated system.
I don’t even think there’s an ounce of anything new here to anyone who has been paying attention.
And I think google offers to translate text from picture for a while? So why wouldn’t they implement this to YouTube? And of course they have to moderate the content if possible.
