At the Consumer Electronics Show 2019, a British company unveiled a service dedicated to VOD solutions, able to detect whether the sharing of a login password is done legally or fraudulently. At the heart of this system, machine learning.
“Can you share your Netflix username and password with your family friends?”. it’s an issue Netflix has long struggled with. Netflix account sharing is very simple, and it’s easy to get around the limitations of the popular video-on-demand service. To counter this shortfall estimated at $10 billion for 2021 by Parks Associates and what amounts to piracy, a company specialized in artificial intelligence has offered its services to Netflix, but also to other services involved in the subscription-based offering video content. At the heart of this anti-piracy system, called Credentials Sharing Insight: artificial intelligence!
The company in question, Synamedia, has developed a software capable of monitoring, detecting and blocking the sharing of usernames and passwords. Why is this solution better than Netflix and competing in-house systems? Simply because the artificial intelligence, according to its creators, would be able to know if the account is shared with a relative in the family circle, or if the credentials are being used by a person who is not related to the subscription. An intelligent monitoring system that avoids unwanted blockages, while identifying fraudulent uses.
To make the distinction, the software first recovers the entire database of the service involved, or tens of millions of accounts when it comes to Netflix. Then, the artificial intelligence comes into play to identify the connection locations and devices used by an account. Whenever the credentials are used at another connection location or device, it is the machine learning that is responsible for determining trends, and giving a probability score. The higher this score, the more obvious the piracy. This is even more obvious when the same account is connected to several places at the same time.
But why use machine learning rather than a tighter blocking system like other services? Simply because user habits change, especially those of young people, more inclined to start a series on their computer, then continue to watch an episode on their smartphone, and finish on their console at a friend’s house. IP addresses are multiplying and it becomes complicated to know if the use is fraudulent or not. An in-house algorithm will not distinguish between the different scenarios while a score of probability, given by the artificial intelligence, leaves the final decision to humans to determine the outcome.
If the account is found to be violating Netflix’s terms, two solutions will be presented: either the account is blocked because it is used illegally; either the user is oriented towards an offer more adapted to his or her type of use. This tool is therefore not only an anti-piracy system since the goal is also to evaluate the different types of content sharing to guide the user to a more complete solution and therefore… often more expensive.