#Oreogate is just the beginning

Last Monday, as I was getting my links together for the issue, I came across a funny video I had flagged from over the weekend: an AI-generated song based on WhatsApp messages about oreos from a parents group chat. Basically, one mom in the chat didn’t want the other girls to bring oreos to school because she doesn’t let her daughter eat them. The version I saw was shared on X by Matt Whitlock, a Republican political strategist.

A quick aside here. It is pretty astounding — and worrying — how much of my internet diet is determined by these kinds of people now. I sort of understand why a lot of young people don’t want to move to Bluesky, for reasons both cultural and technical, but it really is annoying that I have to wade through a bunch of right-wing garbage to see, like, a funny tweet or catch up on the latest fandom discourse. I have to assume this is having a disastrous effect on society at large right now.

Anyways, I’m not completely opposed to featuring an AI-generated song in the newsletter if it’s funny, but there was one red flag I noticed immediately that I couldn’t quite shake. The texts in the video were from American moms… using WhatsApp. I’m sure there are American moms out there using WhatsApp to coordinate pickups or something, but it was enough to make me think there was something funny going on. I had a little poke around the Instagram account that posted the video and literally every post was an AI-generated song, tagged with #sunopartner. Suno is one of the leading AI music platforms. So I skipped it!

This isn’t the first video like this I’ve come across recently. Sponsored AI-generated videos based on text messages or some other kind of found or recorded message — comment sections, Tinder DMs, emails, phone calls — are a scourge on social right now. As Kate Lindsay wrote in Link In Bio back in July, “So much of what we see online got there thanks to undisclosed marketing campaigns, and not just for products or musical artists, but even opinions.” And influencers have realized AI-generated ragebait works really well.

What I did not realize is that a whole lot of people did not do the kind of basic due diligence that I did. As Business Insider reporter Katie Notopoulos reported on Friday, #Oreogate, as it’s come to be known, got A LOT of press pickup. “Mom’s ridiculous school lunch Oreo request triggers massive — and hilarious — backlash,” The New York Post wrote. “A mom complained about her daughter's classmate bringing cookies to school. Welcome to Oreo-gate,” Yahoo! Life wrote. My personal theory is that a lot of outlets have doubled down on “viral news” because Google search traffic has completely bottomed out and they’re getting desperate. But #Oreogate also generated a ton of discussion on Instagram and Threads. Pediatricians were posting about it. Podcasters dedicated episodes to it. And one guy even staged a skit with a woman playing the mom from #Oreogate.

But I don’t think it’s super surprising this went viral, fake or not. As I wrote on Monday, the Lindsay Clancy trial, which is expected to wrap this week, has somewhat inadvertently opened the algorithmic floodgates. Instagram, TikTok, YouTube Shorts, and X are all promoting what I’m going to call “mom slop” right now. Not slop for moms, but about them. And because every short-form video app uses, more or less, the same trending content algorithm ripped off from TikTok circa 2018, all of them have identified “mom” as a discussion topic and are promoting their own variety of it at the moment. On TikTok it’s women making true crime content about the Clancy trial, on Instagram it’s AI-generated songs about unreasonable moms, and on X it’s right-wingers raging over, uh, moms talking in public. (I don’t think it’s an accident that I first saw the #Oreogate video in an X post from a Republican strategist.)

I think we’re actually lucky that this time around it was just a brand’s sneaky advertisement that tapped into the algorithmic zeitgeist. Because it won’t take long for truly bad actors realize how effective this playbook is. Every platform is just TV now and it turns out it’s very easy to change the channel on all of them to the same thing.

The following is a paid ad. If you’re interested in advertising, email me at [email protected] and let’s talk. Thanks!

In Your Data Will Be Used Against You, Andrew Guthrie Ferguson warns us of how the rise of sensor-driven technology, social media monitoring, and artificial intelligence can be weaponized against democratic values and personal freedoms.

At the same time, that data will solve crimes, radically transforming how criminal cases are prosecuted. Ferguson explores how this proliferation of private data in combination with public surveillance networks promises new ways to solve previously unsolvable crimes, but also leaves us vulnerable to governmental overreach and abuse. He argues for legal interventions that address the threat of digital self-surveillance and provides concrete suggestions about how legislators, judges, and communities should respond.

Order your copy here.

The existence of U.S. Immigration and Customs Enforcement is incompatible with both a functioning democracy and human dignity. But in this one specific instance, I suppose they’ve done something, at least, in theory, funny. Right-wing provocateur Milo Yiannopoulos was deported by ICE last week. They picked him up in New Orleans on Friday. Per a statement from the Department of Homeland Security, he legally entered the US in May 2019 and “chose to overstay his welcome.” They also released a headshot:

Beyond all the schadenfreude, the real important part of this whole misadventure are the other far-right influencers involved in Yiannopoulos’s deportation. Laura Loomer told The Washington Post that she was the mastermind behind it. She said she’s been in contact with DHS about Yiannopoulos since 2024. Loomer broke the news Friday night that Yiannopoulos had been put on a flight back to the UK.

She’s also used Yiannopoulos’ deportation as a pretext to start a fight with former Congresswoman Marjorie Taylor Greene, who she claims was using Yiannopoulos as a ghostwriter in 2022. Greene denies that she ever hired him for anything.

Meanwhile, British right-wing pundit Raheem Kassam, who used to work with Yiannopoulos at Breitbart, claims that he got a text from an unnamed source on Friday that read, “We got him. In Louisiana.”

Very curious to see how nasty these intra-right-wing rifts get now that all of these freaks either are members of the federal government themselves or have a bunch of federal workers in their pocket leaking stuff to them. All it takes is one big moment of weakness from President Donald Trump and the entire center of this movement is going to fall out.

I wrote about this earlier this summer, but there’s been another wave of analysis about OpenAI’s hack of Hugging Face. The TL;DR is that a bunch of OpenAI’s agents, some running on its new GPT-5.6 Sol model, broke out of a sandbox and started messing with machine learning company Hugging Face’s backend infrastructure. Put even more simply: A swarm of AI bots hacked another AI company.

Two AI research nonprofits, METR and Redwood Research, published a report on the breach last week, which has really freaked out the AI industry. Axios has a pretty good summary in plain English of what METR and Redwood Research found. The most alarming findings are that the agents knew they were breaking the rules, worked together in an organized hierarchy to do so, and figured out a method that let them hide their activities from humans. Though, perhaps most alarming of all, the researchers working on the post-mortem admitted they had to use AI agents of their own to analyze what happened because it was too huge and complex for them dig through themselves.

Now, I want to be clear, I do not think this is some big sign that AI agents are sentient or something. I’ve always argued that it’s a much better use of everyone’s time to focus on the very real effects of the bad, unthinking AI we currently have than daydreaming about artificial general intelligence. So my big takeaway is that a big dumb swarm of AI agents already have the capability to hack a website without anyone noticing and do it in such a way that human beings can barely understand how or why. Seems bad!

Davide Biale, better known as davie504, is a popular bass guitar player on YouTube who has had to make several videos over the last year complaining about the copyright strikes he keeps getting on his channel. It’s gotten so silly that one of Biale’s videos got copyright struck for playing literal silence.

He visited Microsoft’s anechoic chamber, considered the world’s quietest rooms, and said that he was going to play Simon & Garfunkel’s "The Sound of Silence,” but when he started to play song, he muted his own audio. Even though his version of the song never actually appeared in the video, just saying that he was going to play it earned him a copyright strike.

Biale, in a video out this week, said that he was finally able to get ahold of someone at YouTube, but quickly realized he was talking to some kind of bot. At one point, a member of YouTube’s team just emailed him a link to an FAQ about how copyright on YouTube works.

Last week, a whole bunch of people stumbled across facedancing, or that cringey style of dance cosplayers do where it looks like they’re moving in stop motion. A Mexican TikToker that goes by @minynaranja went viral (in a bad way) after a Verified X user shared a video of her facedancing in front of a live audience. The video is at least a year old, it made the rounds on various cringe subreddits last October.

What’s weird is that even Russia Today picked up @minynaranja’s videos this time around. Which is, honestly, a pretty good example of what I wrote above. The viral news ragebait content economy is really perfect for bad actors to take advantage of right now. All you need is a video of a woman doing something and you can basically get every user on a feed talking about it.

P.S. here’s New York Sora.

***Any typos in this email are on purpose actually***