Nobody Knows How to Stop the Internet's Overzealous Copyright Robocops

The bots that are meant to police sites like YouTube and Ustream for violations of copyright are on a rampage and stirring up trouble across the Internet. Over the course of the past month or so, the streaming of a number of major broadcast events has been derailed by these so-called “algorithmic copyright cops”, attempting to legislate over an unprecedented number of copyright claims. And the humans involved appear to have no idea how to stop them.

Though the incidents happened in different venues, each of them bears the same telltale signs of an overactive copyright apparatus ill prepared for. First, as we reported last month, a clip of the Curiosity rover landing on Mars posted by NASA on its very own YouTube channel was brought down with the message that the “video contains content from Scripps Local News, who has blocked it on copyright grounds.” The footage was all public domain, and even NASA “scolded” YouTube for its oversight. Then, last week, the livestream for the Hugo awards for science fiction was derailed from the Internet, like something out of a bad sci-fi story. After airing pre-approved clips from shows like Doctor Who and Community the feed cut out and didn’t come back. Never mind the fact that the studios themselves had supplied the clips with explicit permission that they could be aired.

The last straw came on Tuesday, when the feed from Michelle Obama’s speech at the Democratic National Convention inexplicably disappeared from YouTube. Again, all signs pointed to a copyright-related takedown. But who owned the copyright?

Well, based on the way these copyright bots work, it doesn’t necessarily matter who owns the copyright or what’s at stake. The first line of defense for these services is almost always the off switch. A number of companies including Audible Magic, Gracenote and Vobile are pushing beyond the requirements of the 1998 Digital Millenium Copyright Act and preemptively patrolling for violations. The DMCA offers a process for reporting and evaluating potential copyright violations as well as justifying takedowns when necessary, but it doesn’t require sites like YouTube and Ustream to pull down content whenever they think it might be in violation.

This, says Parker Higgins of the Electronic Frontier Foundation, is just Hollywood getting these video sites to do their bidding. "The companies that are selling these automated takedown systems are really going above and beyond the requirements set for them in the DMCA, and as a result are favoring the interests of a handful of legacy media operators over the free-speech interest of the public," Higgins told Wired. Even the Dept. of Homeland Security is apparently making claims
to take down content it doesn’t approve.

In the same year that the Internet came out vociferously against iron-fisted legislation like the Stop Online Piracy Act, the erroneous takedowns are in step with a giant leap in copyright claims. In August of last year, the number of takedown requests received by Google numbered 131,577 each week. This August, the number hit a whopping 1.5 million URLs per week, double what it had been in July and an increase of 1,137% over last August. And 37% of all DMCA notices Google receives, the company recently reported, are not valid copyright claims.

What’s even more unnerving about the recent string of takedowns is that even the video hosting sites don’t seem to know what’s going on. Ustream co-founder and CEO Brad Hunstable struggled with an explanation — though not with an apology — in the wake of the Hugo Awards takedown. But he admitted that they were “were not able to lift the ban before the broadcast ended” in an official blog post. He added that Ustream had “suspended use of this third-party system until we are able to recalibrate the settings so that we can better balance the needs of broadcasters, viewers, and copyright holders.”

So these third-party systems must know what’s going on, right? Vobile, the company that’s in charge of policing Ustream for potential copyright violations, brushed off blame and even denied that it had the ability to shut down videos. Its service can only identify the potentially infringing content, Vobile founder and CEO Yangbin Wang told Slate. It’s up to Ustream what happens next. “This was a total surprise to us,” Wang said. “The fact is it’s not true. We were not a reason, we were not a cause to shut down the stream. We have no way to control their system or shut down their channel. We’re not a cop.”

Regardless of who’s to blame for the takedowns, and whatever direction the tide is turning along the dirty beach of digital copyright law, it’s obvious that the bots aren’t cutting it. As the EFF pointed out in a blog post on Friday, these algorithmically powered tools simply aren’t equipped to comprehend the nuances of fair use. Whether it’s clips from shows that have explicit permission to use the content or video in the public domain that everybody has the permission to use, the bots simply match the video and audio of what’s being played against what’s already copyrighted and block the supposedly infringing content at will.

So what can be done? Well, for one, the bots could be reprogrammed, but that’s likely to have limited success. Copyright claimants, who are already “under penalty of perjury” when submitting their complaints, could be charged a fee for submitting erroneous or false claims. Otherwise, YouTube and Ustream could consider using human beings who do understand the nuances of fair use to help police their sites. Humans are already busy moderating these sites for abusive content — why can’t they throw potentially infringing content into the mix and dial back the trigger happy bots? Probably because they’re afraid of getting sued, that’s why.