Facebook App Developer Kogan Defends His Actions With User Data

The app developer who surreptitiously gathered and shared 50 million Facebook user profiles says the company was officially notified of his actions but failed to stop it, according to Bloomberg.

Aleksandr Kogan, a research associate in the department of psychology at the University of Cambridge, turned over his Facebook-generated personality research to the political consulting firm Cambridge Analytica. In an email to university colleagues he called Facebook’s side of the story a "fabrication.

He said that in 2014 he used an official Facebook platform for developers to change the terms and conditions of his app from “research“ to “commercial use,“ and that at no point then did the social media company object. Kogan’s position contradicts Facebook’s stance that he violated the company’s terms and services and then lied about it.

Facebook says that at the time Kogan launched his commercial app there were privacy rules in place that clearly stated it was against company policies to transfer user data externally once obtained, and to use friends’ data for anything other than measuring internal user experiences. In 2015 Facebook discovered that Kogan’s app, a personality quiz, had obtained direct access to 270,000 user profiles and in turn millions more friends who didn’t realize their default privacy settings had allowed the app to harvest their personal information.

Kogan passed the information to Cambridge Analytica, the data firm that helped Donald Trump win the U.S. presidential election in 2016. When Facebook found out about external transfer of the data, it removed his app from the site and asked Kogan to delete the data. Facebook reportedly understood that the researcher complied. A series of news reports called into question whether he removed the data. Kogan says he did.

The app recruited Facebook users through a program called Qualtrics and then asked them to complete a series of surveys. Kogan says that the personality profiles that he gathered ended up not being particularly useful for making predictions needed for micro-targeting.