Sunday April 16, 2017

Potentially "Undefeatable" Ad Blocker Looks at Content, Not Code

Princeton and Stanford Researchers have designed an ad blocker that does its work through computer vision rather than source code. The aptly named Chrome extension, Perceptual Ad Blocker, analyzes words and pictures rather than HTML markup and tags to filter out advertisements. The team claims that the ad blocker remained undetected on all of the sites they tested it on.

News Image

Existing blockers scour a webpage's source code for signs that they are ads, but those can easily be disguised by anti ad-blocking sites. Narayanan's team designed Perceptual to ignore hidden HTML markup (or tags) and look instead at the actual content (i.e., words and pictures on the page). Narayanan and Princeton undergraduate Grant Storey already released a Facebook-specific version of its ad-highlighting extension in November, after the social network announced it would make its ads look like regular posts to thwart blockers. That extension already supports several thousand users, said Narayanan. Today's release extends that ability to the rest of the Internet, targeting AdChoice displays.