Another Copyright Owner Doesn’t Like 512(c)…and Thinks an Anti-Copying Filter is Copyright Infringing–Scott v. Scribd
By Eric Goldman
I hadn’t planned to blog on this lawsuit, but I was finally prompted to read the complaint and thought there was enough stuff to ridicule to warrant a brief post:
There can be a fine line between brilliance and crazy when it comes to new lawsuits, and this one falls on the crazy side of that equation. It is a frontal assault on the DMCA online safe harbors and a backdoor assault on the burgeoning efforts to develop anti-copying filters. I think both efforts are wildly misdirected.
The 512(c) Assault
With respect to 512(c) assault, the plaintiff doesn’t bring much new to the party. If 17 USC 512(c) preempts all flavors of copyright infringement (both direct and secondary), then this lawsuit is probably DOA.
Note: Fred kindly linked to my write-up of the recent UMG v. Veoh ruling but took issue with my assessment that it’s ambiguous if 512(c) only preempts direct copyright infringement or all flavors of infringement. He correctly notes that the legislative history is entirely clear that 512 preempts all flavors…but this only helps if a court reaches that legislative history. On its face, the statute has express exclusions that mirror language from the secondary copyright infringement tests, so other courts could simply rely on the statute to narrow the statutory protection only to direct infringement. Indeed, a number of courts have implicitly reached this interpretation without acknowledging the ambiguity; which made the UMG v. Veoh’s determination that the words in the statute mean something different than the exact same words used as part of the vicarious copyright infringement test all the more interesting.
Back to Scribd’s 512(c) defense. Recall that I said the lawsuit was “probably” DOA. As part of trying to oust Scribd from 512(c), the plaintiffs challenge Scribd’s compliance with the statutory formalities by noting an inconsistency between Scribd’s contact info on its site and in its copyright office filing. (See para. 35-36). It’s possible that this highly technical “problem” could be a greater threat to Scribd’s 512(c) eligibility than the plaintiff’s more substantive assaults.
The Copyright Filter Assault
With respect to the fact that Scribd uses copyrighted works to create a filter to prevent their reloading, I think this lawsuit is way, way off-base.
First, the Fourth Circuit recently held that the Turnitin anti-plagiarism filter (built using copyrighted works that were, in some cases, included in the system through quasi-coercion) was a fair use. See my blog post on the district court ruling AV v. iParadigms for more detail on the coercion and fair use points. Further, in that case, Turnitin monetizes its database directly by licensing access to it, while the complaint apparently alleges that Scribd uses the database only for its own filtering purposes and does not license it to others. As a result, Scribd’s internal usage of the database (without separate monetization) seems like it would tilt the fair use equitable factors even more clearly in its favor.
Alternatively, courts could reach fair use by concluding that Scribd’s copying into the filter is an intermediate copy to create a non-infringing/fair use output. See, e.g., Sega v. Accolade; Kelly v. Arriba; Perfect 10 v. Amazon. In fact, as far as the complaint indicates, Scribd never republishes any of its copy used for filtering, while Turnitin republished portions of its copies as part of its originality reports.
Second, several courts recently have encouraged defendants to build filters to suppress infringing or wrongful behavior and given the defendants extra legal protection because they are using these filters. See Io v. Veoh; UMG v. Veoh; Tiffany v. eBay; Cisneros v. Yahoo. For example, the recent UMG v. Veoh case lauded Veoh for upgrading its infringement suppression filter from a weak hashing technology to the more robust Audible Magic system. Given this trend, it seems likely that future courts will legally protect efforts to build more robust anti-copying filters, even if copyrighted works are copied along the way.
Third, this attack on the filtering efforts, and the fact that it’s probably not Rule 11 sanctionable despite its ridiculousness, shows just how misdirected copyright law has gotten. The plaintiffs seek to set up what Wikipedia says is a “Morton’s Fork” (which, I think, most people refer to as a Hobson’s choice). A service provider can choose not to automatically filter the site, in which case infringement can run rampant and excise the service provider from 512(c) based on the statutory technicalities and exclusions. But if the service provider builds a filter to automatically filter for copyright, the plaintiffs believe the service provider should face liability for building the filter.
Courts recognize and would like to avoid this dilemma, and the copyright owners’ countervailing interests are weak. Other than the theoretical and highly formalistic loss of control over their copyrighted works, exactly how are copyright owners harmed by having their works in Scribd’s anti-copying filtering database? (Some of us would argue that the only harm is if the filter actually blocks potentially valuable exposure that might have come from having the works published through Scribd, but I don’t think that’s what the copyright owners are complaining about!). The filter helps copyright owners prevent future infringements, so in that respect it only preserves, not reduces, the value of the copyrighted works. Thus, without any harm, lawsuits like this simply reflect a ridiculous effort to use copyright law to stop progress. This isn’t what copyright law is about, and it’s dispiriting to see some copyright owners try nonetheless.