👁️ Instagram's Fake People Are a Catastrophe (for Meta)
Also: Do Kwon is Coming to America! PayPal's giant click-scam, FTX updates, and more.
Happy New Year! It’s time for more hilariously embarrassing efforts to turn AI into a product. But first, we’ve got scams from PayPal! Scams from Ohio police! Do Kwon updates! And more!
The Honey Plot
In one of the most explosive, crazy scam busts in recent memory, YouTube investigator MegaLag has discovered that the PayPal-owned “discount code search” extension Honey has been a massive fraud for years. Very briefly, the extension pretends to search for coupons to front-run other referrers to get the “last click” before a transaction, which is how referral fees are calculated. At the same time, they’ve been cutting deals with retailers to control which discount codes it actually displays - making its many claims to “find all the best discounts” manifestly fraudulent.
This is going to court, and it’s going to result in a *gigantic*, complicated settlement.
Ryan Salame Knew FTX Was a Fraud
New claims in a civil lawsuit seem to overturn claims of ignorance by FTX’s second-biggest whiner, Ryan Salame. Messages suggest he not only knew about Alameda’s back-door borrowing of customer funds, but provided advice about how to make it seem more legitimate.
Tether Invests $775m in Rumble
The offshore stablecoin issuer throws off cash like a printing press, and it has to go somewhere. Why not curry favor with the incoming administration by wildly overvaluing a right-wing media platform, whose investors have included J.D. Vance and Peter Thiel?
Do Kwon Extradited to the U.S.
Stanford-educated Do Kwon is even more of a paradigmatic “smart dumb guy” than Sam Bankman-Fried. He really believed his financial perpetual motion machine would work! But finally, after about a year, he’s being extradited from Montenegro, where he was stopped attempting to flee to the Emirates.
What I hope for most is that we get a trial at SDNY. Please, Do, plead innocent! You’re innocent, we believe in you! You can win!
Congestion Pricing: u mad bro?
After useless Kathy Hochul tried to slow-walk it, congestion traffic pricing went into effect in Manhattan yesterday. Evidence is still very limited (weather was bad), but traffic looks positively blissful out there. I don’t drive, so I’m biased, but endlessly entertained by the transferred rage manifesting from right-wingers everywhere at the very idea of managing a public good for the public benefit. For some perspective, it now costs $9 each way to drive into Manhattan. What hasn’t changed is that it costs about $50 a day to park in Manhattan. Cry more.
The Price of Your Own Murder
Ohio Governor Mike DeWine has signed a law allowing law enforcement agencies to charge hundreds of dollars for body cam footage, putting up a substantial barrier to investigating America’s continuing plague of police misconduct/extrajudicial killing.
ProPublica’s Militia Mole
Speaking of which, a new longform report from ProPublica details a hare-brained one-man undercover operation to uncover white supremacist cops. A seemingly novice spy decided to go undercover for more than two years with the Oath Keepers and other militia groups, doxx them extensively, then release his findings.
Juicero is Still Captivating
The infamous epitome of startup fraud is going viral again. The company - which had more than 230 employees in 2017! - was laid low by a hard-hitting Bloomberg investigation that discovered the $699 Juicero machine wasn’t actually doing anything. You could just squeeze the bags by hand. The juice was just in the bags. This company raised $120 million dollars. Incredible.
The Momfluencer to Child Trafficking Pipeline
TW: Grim Stuff. Following a particularly weird lifestyle reel going viral, this disturbing thread breaks down how “momfluencers” on social media are incentivized to sexualize their own children. It’s horrific.
Facebook’s Fake AI Users Get The Boot
Did Facebook re-hire David Marcus? They’re sure fucking up as if he’s in the room!
Here’s the latest inept own-goal from a company that hasn’t succeeded at anything new in nearly a decade: The day after Christmas, the FT published a piece about Meta/Facebook’s plan to let horrible AI bots (they’re not agents, they have no agency) run rampant on their platforms.
“We expect these AIs to, over time, exist on our platforms, kind of in the same way users do,” ran one staggering quote. “They’ll have bios and profile pictures and be able to generate and share content powered by AI on the platform.”
Pretty soon after, people spotted some really corny bots like Grandpa Brian (which are both stupid and, I gotta say, a little offensive) and thought Facebook was already launching their agents. The reaction was, people were so incredibly pissed off that Meta/Facebook had to make clear that these were old profiles that were part of a previous test.
The gigantic company run by dimwits had to frantically backtrack and delete the older fake people, which they had forgotten to do (sure, Jan) nine months earlier, presumably to make room for new, better fake people that users won’t loathe quite as viscerally because …
Well, that’s where they’re turbofukt, actually. Because Mark Zuckerberg has already announced to investors that letting AI run wild is their new strategy for Juicero-ing engagement on their platforms. And yeah there’s some slice of delirious boomers whose kids have cut off contact, who will keep clicking on whatever you put in front of their collapsing faces.
But more people than Zuck seems to believe can tell when something is fake. And they don’t like it! I mean Christ, if you’re creating fake people, maybe you could at least make them more interesting than this swill? Oh but you can’t because LLMs are literally built out of averages. Nobody wants to feel like they’re being pandered to, and that’s built into the technology.
Sure, AI sucks, but this isn’t about Meta inflicting some horror on a captured pool of users: it’s about the last desperate gasps of an entire social media economy, which Meta/Facebook in 15 years has repeatedly failed to grow beyond or even meaningfully improve. They’ll do some damage on the way down (just like they did on the way up) - but there’s only one direction from here.
Getting AI bots to gin up fake engagement that generates a hollow ache in user’s chests won’t work any more than “The Metaverse” to reverse Facebook’s steadily declining fortunes. Notably, even Morning Brew, which is quite a pro-tech outlet, referenced dead internet theory in their tweet about the fake users. While I take an obvious moral stance about these things, I’m also a business reporter by trade. I don’t think this is a “take major shorts against Meta” thing, because that would be crazy. But it seems very clear that if they’re chasing the AI/LLM dragon, they’re entirely out of ideas.
Mark Zuckerberg just turned 40. That means he’s going to get to spend half of his life as a rich guy who invented something that was successful for 15 years and then shambled on as an embarrassing shadow of itself. You could do worse, obviously, but if I were in his shoes I wouldn’t be selling my eternal reputation for a mess of AI-generated pottage.
He controls the company. He doesn’t need more money. He’s doing this because he’s trapped, and not actually very smart! He chased crypto and that failed. He chased “the metaverse” so hard he changed the name of the company … and that failed, spectacularly. Now he’s chasing AI with an objectively terrible implementation. It’s already failing.