In 2014, when the current artificial intelligence boom was still in its infancy, a casual exchange between tech executives set the stage for a decade-long privacy dispute. Matthew Zeiler, founder of the AI startup Clarifai, emailed OkCupid co-founder Maxwell Krohn with a blunt observation: the dating site possessed a "HUGE amount of awesome data" for facial recognition training. Krohn complied, handing over three million profile photos—a transaction that directly violated OkCupid’s own privacy policy.
The details of this arrangement only surfaced years later, following a 2019 investigation by *The New York Times* that prompted the Federal Trade Commission to intervene. Clarifai had used the images to hone a facial recognition service capable of identifying demographic markers such as age, gender, and race. It was a classic example of the "move fast and break things" era of data acquisition, where the personal intimacy of a dating profile was treated as raw industrial fuel.
Last month, following a settlement between the FTC and OkCupid’s parent company, Match Group, the consequences finally caught up with Clarifai. The Delaware-based firm has now certified to the FTC that it has deleted the three million photos. Crucially, the company also confirmed to Representative Lori Trahan (D-MA) that it has destroyed the algorithmic models trained on that specific data. This "algorithmic disgorgement" is an increasingly common tool for regulators, ensuring that companies cannot continue to profit from the mathematical ghosts of illicitly obtained data.
With reporting from Engadget.
Source · Engadget
