Y’all, someone done went and Groked Substack’s video-editing AI

The first time I noticed it, a listener told me about it. On a video where I’d recommended boycotting Chase bank, the video transcript inserted the word “support” beside Chase bank. Odd, I thought, but a one-off.

The exact same listener reached out again to tell me about another instance transcript sabotage, where Substack’s video-editing AI inserted a word I never said so that the transcript read “I am glad CSA happened”— only I never said the word “happened.” AI inserted it into the transcript, effectively negating my central point. That sneaky AI “edit” took place in a post where I was critiquing what is clearly a billionaire astroturf messaging campaign to claw back the age of consent when it comes to teens, in response to the Epstein files fallout. Disgusting stuff.

Tonight, Substack’s AI video-editing software didn’t insert a word I didn’t say— it straight up airbrushed out of my video two different critiques I made of MAGA politicians: (1) where I critiqued Mike Johnson for saying the Epstein files couldn’t be released unredacted because of “national security” and (2) where I called out the Trump regime’s war of aggression against Venezuela— which, once more, has nothing to do with “national security” and everything to do about mobster imperialism. Both of those points were entirely removed by AI from the original post.

This is some Grok-ass b******t. And it’s a real problem, because the AI features are set to ON by default.

So here are some calls to action:

* If you’re a listener and you notice weird transcript discrepancies, or wholesale deleted portions of someone’s Substack live recording, message them about it.

* If you’re a Substack video creator, consider turning off the AI features, which are set to default. In the top lefthand corner of your video, you’ll see an AI starburst that says “Edit AI Features.” Click on that. You’ll notice two tabs are auto-selected. Unselected both and hit save.

* If you’re someone like me who does Substack lives, don’t immediately publish the video. You’re probably going to want to go back and check to make sure the AI video function doesn’t hijack your message.

* If you begin to notice a pattern, think it might be important to document this stuff, because sneaking in censorship through AI video editing ain’t okay.

What do I think is going on here? Y’all, computers ain’t my ministry, but I do know that if Elon’s racist ass is trying to make his AI more Nazi-like, it’s probably possible there’s some techno-fascist key word key, and they’re using AI like a free-speech roomba and “tidying up” all the speech that hurts billionaire fee-fees.

Either way, it’s b******t. Be on alert.

This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit muppetgender.substack.com