Clearview AI is the Weirdest Company in the World
Ad Hominem #7 - Machine Learning, Corporate Malfeasance, and Congressional Overreach.
I want to focus some attention on one company, Clearview AI (company website), which has been popping up in the news recently (New York Times, Vox). Clearview AI recently had its entire customer list stolen (The Daily Beast, Izod News, The Hill) and Apple has suspended the app distributed by the company to its clients because it was incorrectly using Apple’s app-sharing tools for corporate developers (Patently Apple, BuzzFeed). While these stories are notable, there is a much bigger story here about the company’s business goals and practices.
Clearview claims that they are not a surveillance system, despite the fact that they are now making security cameras and looking into making augmented reality tools through a sister company (BuzzFeed). This piece from Vice goes in deep on just how much information they’ve gathered, and how difficult it is to get them to give it up (Vice News).
They have amassed their massive facial database (3 Bn+ images) by pulling down public images from blogs, Facebook, Twitter, Google Photos, and any other source they can find, often with spooky findings from deep web archives. This has caused Google, YouTube, LinkedIn, and other companies to send cease-and-desist letters to the company, thus far ineffectively (Engadget, CBS News). Meanwhile, the company is moving beyond public images and trying to acquire all US mugshot photos from the last 15 years (OneZero).
This piece from CNN features an interview with Clearview CEO and Founder, Hoan Ton-That:
For more, watch PBS’ interview with the founder (PBS).
Bits of News:
💰The FCC has proposed $200M in fines for wireless carriers who have been selling customer data for years (TechCrunch). This is just another slap on the corporate wrist with no real lessons learned or criminal penalties earned. Selling the data of ordinary Americans to the highest bidder, of course, is not a new practice but rather one that has been going on for decades (Saturday Evening Post).
🧠This video from Think With Google, their business advertising thought leadership site, features Sir Martin Sorrell spreading some euphemistic manure about personalized digital marketing. Worth a watch only if you’re curious to witness the banality of the discussion around carving up humanity’s time, attention, and money. One key innovation he touts is the invention of the 2-second ad insert on mobile. When I was a kid we just called those pop-ups (Think With Google).
🔐One important factor in privacy is of course the ability for people to communicate over encrypted channels. This basic property of internet communication has been under threat for decades, but now the “EARN IT” Act means it’s time for this same old fight again (Cryptography Engineering, TechCrunch).
💎Twitter has a new “Stories” product called “Fleets”. We can’t ban the Nazis, but I guess we can make their posts disappear after 24 hours (TechCrunch).
☎️Comcast “accidentally” shared and published 200,000 phone numbers that its customers had been paying to keep unlisted (Ars Technica).
👕The New Yorker has a piece on adversarial fashion that seeks to undermine AI surveillance systems (The New Yorker).
👮♂️Live facial recognition is coming to police body cameras (OneZero).
Ad Hominem is a newsletter about Ad Tech, Privacy, and Surveillance Capitalism. It currently has no paid subscription. If you want to support this project, the best way is just to spread the word via Twitter or forwarding this email to a friend. You can also send me a tip on my BuyMeACoffee page.
If you have any feedback on this or any other issue of Ad Hominem, please let me know on Twitter. Thank you for your time and attention. I know it’s your most precious resource.