Slashdot Mirror
Video Services May Use AI To Crack Down on Password Sharing (variety.com)
An anonymous reader shares a report: Still using your ex-roommates cable credentials to watch "Game of Thrones?" That may soon be getting a lot harder, thanks to new efforts to crack down on password sharing for pay TV and online video services. One of these efforts, launched by London-based Synamedia ahead of next week's Consumer Electronics Show (CES), even uses artificial intelligence to uncover notorious password sharers. Credentials Sharing Insight, as the new service is being called, targets both casual password sharing as well as criminal enterprises looking to resell pay TV login information. However, the focus clearly is on friends and family taking their generosity a bit too far, explained Symanedia chief product officer Jean-Marc Racine in an interview with Variety this week.
[...] Most services have tried to curtail password sharing by limiting the number of simultaneous streams, with little else to go by to identify abuse. "Today, you are in the dark," he said. Synamedia's solution on the other hand digs through lots of data to cluster users based on their streaming behavior. This can include user's physical location (someone streaming from both coasts at the same time) as well as general usage patterns (someone streaming 24/7). The company can even take a look at the specific content streamed by a user to identify unusual patterns. Based on these clues, Synamedia trains models to score users on a scale of 1 to 10, indicating whether they are likely sharing their passwords or not.
[...] Most services have tried to curtail password sharing by limiting the number of simultaneous streams, with little else to go by to identify abuse. "Today, you are in the dark," he said. Synamedia's solution on the other hand digs through lots of data to cluster users based on their streaming behavior. This can include user's physical location (someone streaming from both coasts at the same time) as well as general usage patterns (someone streaming 24/7). The company can even take a look at the specific content streamed by a user to identify unusual patterns. Based on these clues, Synamedia trains models to score users on a scale of 1 to 10, indicating whether they are likely sharing their passwords or not.
Also, none of those metrics are failsafe. Some legitimate customers will travel a lot (varying location), work alternating day and night shifts (stream at seemingly random hours), and/or have irregular sleep cycles or suffer from insomnia (sometimes spending all day and night watching Netflix). Judging peoples taste in series to judge whether they are one or multiple people also sounds like something that can backfire for people that legitimately have unusual tastes, or people that often have friends visiting with different tastes. I hope their solution is to at least ask the customer for an explanation, instead of an "AI" simply autobanning all "weird" customers.
I still share my Netflix with my ex. She's willing to pay for Netflix herself, but there is a problem. She can't take her watching history, ratings, and her list of bookmarked titles to her new account. When asked, Netflix say "meh, just start over".
Come on dudes, your devs could easily add account merges and splits (and while at it, give me an option to let me watch the end credits in full screen in peace). I'm not motivated enough to write a scraper/saver to copy her profile from my account to hers using Greasemonkey (the watch list is probably easy to get, but the entirely to watch progress and the simulation needed to bring every timestamp over might be harder on my end), and you guys don't seem either.
But hey, AI is cool. Decent profile handling is so meh. Don't forget the blockchain.
"Everybody's naked underneath" -- The Doctor