Faked Up 2024 Impact Report

This newsletter led to as many takedowns as it has subscribers

Photo by Miguel Á. Padriñán

Hi folks!

I’m back from the holiday break and to compensate for two weeks of silence, I’ll be sharing three posts over the next three days.

First off, a quick look at some key results from this newsletter in its first seven months of existence. The most obvious impact came in the form of Google and Meta taking down at least¹ 1,000 bad ads that I uncovered. Specifically:

• FU#30 led Google and Meta to take down at least 100 ads for a “feng shui bracelet” that was using photoshopped images of celebrities to target folks desperate for a lucky break.

• FU#29 led Google to block an advertiser that ran ~300 deepfake scam ads for a shady supplement.

• FU#12, FU#16 and FU#17 led Meta and Google to take down more than 600 ads and two apps for AI undresser services.

Findings from FU#12 were also cited in a letter on non consensual deepfake nudes written to Google CEO Sundar Pichai by a bipartisan group of members of Congress.

In researching FU#7, I asked OpenAI about the use of ChatGPT to power a British political chatbot. This was later terminated, but unlike with the cases above, I can’t attribute the takedown to my work for sure as I never heard back from the company.

Less tangibly, but perhaps more systemically, FU#28 contributed widely-cited data to the coverage of BlueSky’s challenges with impersonation (see: Engadget, MIT Technology Review, The Register, AP News, Washington Post). The platform subsequently cracked down on impostor accounts. To be clear, I’m not saying Faked Up led to the policy intervention, which was certainly already in the works, but by giving a catchy data point to media outlets it may have raised the salience of the issue with decision-makers at the company.

I spent many hours in weird corners of the internet for this year, whether it was the Catholic Facebook pages pushing AI slop or X’s Election Integrity Community suddenly disinterested in proving voter fraud. I often know where I’m starting but not where I’ll end, as with this item about a Tesla phone hoax on LinkedIn that led me to find hundreds of deepfake scam ads for a penis enhancement pill.

I’m probably proudest (not sure that’s the right word?) of my deep dive on AI-generated thirst traps on Instagram, which informed this great 404 Media/Wired piece.

As of December 31, Faked Up had 1,287 subscribers, of which 125 were paying. I provide paid access for free to an additional 56 subscribers, most of whom are members of the Break Through Tech program.

Honestly, I thought I would have had more subscribers and less impact by now. This is a trade-off I’m glad to make (for now)! We’ll see what 2025 brings.

I’ll see you again tomorrow with the launch of a new resource I’ve been working on that I hope will be useful to folks reading Faked Up. And then we’re back to regular programming on Wednesday.

Thanks for sticking around,

Alexios

1. Due to the way that both Google and Meta’s ad libraries work, I only have approximate figures. I’m also assuming that investigating my leads with internal tools the companies found additional policy-violating ads.