Google and YouTube have been saying for quite some time that the video-sharing service was working on technology to scan video uploads for material that could infringe copyright. YouTube has now announced the launch of its “Video Identification” tool, albeit in Beta form.

This is a significant development. Commercial copyright owners have long complained that YouTube is awash in infringing videos, and that the DMCA’s “notice-and-takedown” process amounts to a cumbersome and ongoing game of whack-a-mole. They have argued that YouTube should filter out infringing material on the front end — before it gets posted — rather than waiting for content owners to find each infringing posting and send a formal request for it to be taken down. The law doesn’t currently require that, but YouTube’s new tool appears to be a major step in that direction.

YouTube’s announcement leaves a lot of open questions — and the details on the official Google blog and YouTube’s Video Identification page are pretty sketchy. Our understanding is that the system will use some kind of digital fingerprinting technology to check video uploads against a database of copyrighted material submitted by copyright holders. If a match is found, what happens next will depend on whether the copyright holder has asked to block all uploads, allow them for promotional purposes, or “monetize” them (presumably by sharing in revenues from advertising).

From CDT’s viewpoint, the questions that immediately come to mind relate to accuracy and fair use. YouTube is a marvelous platform for free expression, and we wouldn’t want to see legitimate user-created videos misidentified and blocked. And even if the system never wrongly identifies a video clip, what will happen when users incorporate short clips of commercial video content into their own videos, for purposes of criticism or commentary? That’s classic fair use activity. But if the YouTube technology identifies even short clips of copyrighted material within a longer video, such fair use activity could be flagged and perhaps blocked. How the system handles fair use activity, and how users are able to navigate the fair use issues in practice, will be extremely important.

This is a tricky challenge; the case-by-case nature of fair use makes it impossible to translate neatly into an automated process. YouTube should explain to its user community how it will deal with this difficult problem. At a minimum, it seems clear that the video identification system should include some opportunity for users to challenge a positive match on either accuracy or fair use grounds. We understand that the system may in fact include a process for that, but more details are needed. And for that opportunity to be meaningful, users will need some links to balanced information about fair use, so that they can make a reasonably educated judgment about whether their activity is in fact permissible. It also could be useful for the system to include some way to monitor any potential impact on fair use — perhaps a mechanism for users to report cases where they were blocked or dissuaded from posting videos that they believe to be fair use. A longer term question relates to the impact this announcement could have on the evolution of the marketplace and legal regime. The copyright industries would like to see filtering systems become standard industry practice, a strong norm, and perhaps even an affirmative legal responsibility. But not every service or platform is backed by the deep pockets of Google, with the resources to spend months developing and implementing a filtering system custom-tailored to the service’s particular needs. Smaller, innovative services may find that the off-the-shelf options are a poor fit, don’t scale, or are costly. And other services may differ from YouTube in ways that make the challenges and problems associated with filtering quite different.

In short, it would be a big mistake to make a leap from YouTube’s voluntary adoption of a filtering system to any kind of mandatory, affirmative obligation that would apply more broadly. The legal principle that ISPs and platform providers can offer services without becoming content policemen or gatekeepers and without assuming liability for all user behavior has been crucial to the success of the Internet. Mandatory content filtering would be a major departure from this principle. Going forward, policymakers need to be careful about what lessons they draw from YouTube’s example.

This entry was posted on Tuesday, October 16th, 2007 at 5:06 pm and is filed under CDT, Digital Copyright, Free Expression. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.