Content and revenue protection specialist Friend MTS has revealed that the main factor driving development of a third-generation version of its ASiD subscriber watermarking and monitoring solution – demo’d at IBC this year – is that premium content providers are currently witnessing a significant increase in more sophisticated attacks against this type of technology.
Meanwhile, the launch of an OTT version of the new product highlights how this type of piracy is increasingly moving away from linear to online TV delivery, it says, particularly in the live sports field.
Friend MTS is perhaps unusual in being publicly explicit about the precise nature of such attempts to circumvent watermarking, which Jonathan Friend, the company’s CEO, translate into “a significant increase in sophisticated visual manipulation attacks.”
In the past, reveals the company’s director of product marketing, Neil Sharpe, typical attacks might have used techniques such as “masking and cropping” – where the pirates tried to ‘cut out’ the subscriber ID/watermark which can identify their device (this could have been visibly or invisibly present on the screen), or mask the part of the screen image where it’s located, for instance by only showing a cropped version of the screen. This is designed to undermine the watermarking process.
Today’s more sophisticated pirate, according to Friend, uses more advanced techniques such as ‘collusion’. For instance, outputs from four separate set-top boxes can be combined by a pirate in an attempt to confuse the watermark detection system through each one contributing a different quadrant of the screen, each including a different watermark.
Another modern attack-type – specific to Internet video delivery – is ‘OTT player interference’, which attacks the app in the device which plays back the OTT video in an attempt to tamper with the subscriber ID.
Sharpe emphasises that so far, ASiD, which inserts a watermark client-side (which is to say it identifies the actual device pirating the content by invisibly burying data in the video it plays back) has never once been compromised by these or any other attack techniques.
For instance, ASiD’s watermark runs across the whole picture, which makes it difficult for ‘cropping’ techniques to work. Meanwhile, resisting an OTT player interference attack is done by using ASiD’s monitoring system: “What we offer is detection of interference so that when people try to interfere with it, we can identify it and then revoke the licence to the subscriber – so therefore you hold on to the content and it’s not released.”
At the company’s IBC press conference, Friend revealed that some of these new types of threat had been anticipated and protected against in the design of ASiD three years ago. Today’s third-generation version of ASiD had also been built with new types of future attack in mind, he said.
However, ASiD does much more than preventing pirates from concealing their identities: it also provides a global monitoring system to help take them down, which though highly automated, also uses actual human beings for some of the more sensitive work.
“It’s basically cloud-based automated scanning of known sites of piracy apps, and on a huge scale: we download 3Tb of video every day,” notes Sharpe. “It’s just not possible to look at content on that scale with human eyes. [But] humans are there to support the automation and interpret things.”
While Friend MTS is reluctant to use the ‘AI’ word – on so many lips at IBC – it will admit to using Machine Learning techniques as part of its automated processes. “For example, when hunting for [pirated] content, sometimes it can be slightly hidden, and we have processes which allow us to know where to look for it, which incorporate elements of ML.”
However, Friend MTS’s chief marketing officer, Justin Paul, points out that “what we’re doing is looking at people’s real content, paid-for content, and you have to be very careful, particularly when you are using some of the more aggressive counter-measures, switching off people’s set-top boxes and terminating their services. And so there is still that human element. Is this a genuine piracy attack or is it, for example, legitimately streaming something that they have the right to?”