This part is troubling: In May 2014, Dave Willner and Dan Kelmenson, a software engineer at Facebook, patented a 3D-modeling technology[1] for content moderati

This part is troubling: In May 2014, Dave Willner and Dan Kelmenson, a software engineer at Facebook, patented a 3D-modeling technology[1] for content moderation ... First, the model identifies a set of malicious groups - say neo-Nazis, child pornographers, or rape promoters. The model then identifies users who associate with those groups through their online interactions. Next, the model searches for other groups associated with those users and analyzes those groups "based on occurrences of keywords associated with the type of malicious activity and manual verification by experts." This way, companies can identify additional or emerging malicious online activity It's not hard to imagine how this could be misused: Automated guilt-by-association, and within 3 degrees of separation, triggering investigation by authorities (corporate or otherwise). EDIT: And also this tool of automated censorship, though I doubt it will surprise many HN readers PhotoDNA works by processing an image every two milliseconds and is highly accurate. ... Then PhotoDNA extracts from each image a numeric signature that is unique to that image, "like your human DNA is to you." Whenever an image is uploaded, whether to Facebook or Tumblr or Twitter, and so on, he says, "its photoDNA is extracted and compared to our known ... images. Matches are automatically detected by a computer and reported ... for a follow-up investigation" Currently used for child exploitation images, but for what else too and by whom? They add that it's being applied to something more political: Farid is now working with tech companies and nonprofit groups to develop similar technology that will identify extremism and terrorist threats online - whether expressed in speech, image, video, or audio. [1] http://www.google.com.gh/patents/US20080256602 http://www.google.com.gh/patents/US20080256602