• Out of Sight, Out of Mind – The Secret World of Keyword Ads
• Ninth Circuit Reverses Ruling in Perfect 10 v. Google Case

Doubtless only a lawyer could sensibly parse the statement that "commercial use is not the equivalent of ‘use in commerce.'" To other sentient beings, it might seem that the only way of making "commercial use" of something would be "in commerce." Nonetheless, navigating this seeming logical labyrinth, lifted verbatim from a recent decision dismissing a suit concerning internet keyword advertising, may hold the key to what limits henceforth can be placed on such promotional tools. And maybe more importantly, how other more subtle and possibly insidious secret uses of trademarks to guide or divert Internet traffic can be regulated.

Keyword ads are, of course, paid links provided or keyed by Internet search engines to given terms (including trademarks), the ads being then presented with other unsponsored or natural search results concerning those keyword terms. When a car dealer pays to have a search engine such as Google or Yahoo! present a link to its website whenever an Internet user searches generic motoring terms, there is no problem. When the same merchant pays for links to its website whenever someone searches proprietary General Motor–ing terms, the legal question arises whether this is a trademark infringement actionable under the Lanham Act. The courts are now split whether, as a matter of law, such an admittedly commercial use constitutes use in commerce. How this question (and its seemingly paradoxical logical underpinnings) is resolved remains very much in doubt, and will likely impact what restraints, if any, can be placed not only on keyword ads but also on the next generation of Internet marketing strategies.

The case quoted above, Merck & Co. v. MediPlan Health Consulting, Inc., 431 F. Supp.2d 425 (S.D.N.Y. 2006), offered a simple, certain solution to the question whether keyword ads risk infringement liability. That solution, however, will surely do nothing to slow the growth of keyword advertising, not to mention other so-called search engine optimization techniques (about which more in a moment). The mark and keyword was ZOCOR. The defendants paid Google and Yahoo! for keyword links to their on-line pharmacies selling the generic version of ZOCOR. The court reaffirmed its earlier decision dismissing Merck's claims and holding that the keyword linking simply was not a "use in commerce" of Merck's trademark. The court relied on the fact that any use of the trademark involved an unseen automated function that did not publicly use the mark anywhere on or in connection with the goods. In so-holding, the court relied in turn upon another recent ruling involving pop-up ads, which are in many ways similar to keyword ads. In the pop-up ad decision, the Second Circuit held, in essence, that use of trademarks in an automated, machine-linking function, to generate such advertisements was not a use in commerce. 1-800 Contacts, Inc. v. WhenU.com Inc., 414 F.3d 400 (2d Cir. 2005).

In 1-800 Contacts, the plaintiff sued for trademark infringement after WhenU's adware program caused pop-up ads for the plaintiff's competitor, Vision Direct, to appear when Internet viewers visited the plaintiff's website. The Second Circuit (consistent with numerous precedents) held that adware companies do not engage in "use in commerce" within the scope of the Lanham Act when, even for a fee, they employ technical means to cause computer users visiting specific Internet shopping sites to see pop-up ads for competing merchants. 1-800 Contacts turned on the Lanham Act requirement for proving infringement that the defendant has used the challenged mark - or a copy or colorable imitation of it - "in commerce . . . in connection with the sale, offering for sale, distribution, or advertising of any goods or services." 15 U.S.C. § 1114. Relying on the statutory definition of what "use in commerce" means for a claimed trademark owner to establish rights in its mark in the first place, rather than the arguably broader test of infringement (namely, whether a use might be deemed "in connection with … the advertising of goods or services"), the Second Circuit held that the entirely veiled machine-linking function was not a trademark use. The court reasoned that such use did not entail having the mark "placed . . . on the goods or their containers or the displays associated therewith. . . ." 15 U.S.C. § 1127. Part of the same statutory definition of a trademark, which the Court did not address, also provides that a mark is used for services when it is "used or displayed in the sale or advertising of services and the services are rendered in commerce …" (Emphasis added).

In finding no such use in commerce to constitute infringement, 1-800 Contacts correctly explained, in language equally applicable to most keyword ads: "The fact is that WhenU does not reproduce or display 1-800's trademarks at all, nor does it cause the trademarks to be displayed to a [computer]-user." Sponsored keyword ads are essentially no different: they appear alongside native or natural search results, while the keyword search terms and software all function out of sight. And the ads that are displayed (whether pop-ups or keyword ads) typically do not themselves use or display the subject mark or a colorable imitation of it. They are, to the contrary, simply juxtaposed conveniently next to, over, or under the other content of interest. (When a mark is actually used in the visible text of a keyword ad, that presents a simpler case, and a finding of use in commerce is less in doubt. This is true even where the court accepts the basic premise in Merck that sale of keywords does not, in and of itself, entail use in commerce. Hamzik v. Zale Corp., 2007 WL 1174863 (N.D.N.Y 2007).

As this author suggested in a September 2005 article in Managing IP, The Problem Putting Down Pop-ups, 1-800 Contacts and other like cases address only half of the statutory test of what is a trademark (or more specifically a service mark). Those cases do not decide whether or when pop-up ads or keyword ads should be deemed likely to cause confusion, or whether some such uses should be immune from suit under the fair use doctrine. If the question is whether Google or Yahoo! (or the various adware firms promoting pop-ups) "use or display" the subject trademarks "in the sale or advertising of services," and whether those services are then "rendered in commerce," to say the search engines and adware companies do not "display" the trademarks appears to address only half of the statutory test, thus depriving the term "use" of any meaning independent of the term "display." No less puzzling is that the courts have focused on the definition of how a trademark owner establishes rights in the first instance, rather than the definition of infringement. Namely, courts focus on whether the marks sold as keywords or employed to generate pop-ups are used "in connection with … the advertising of goods or services." It surely seems possible a mark can be used "in connection with" advertising even if it is not "displayed" on goods or services.

Although unassailably true that keyword ads (like most pop-ups) typically do not “display” the subject trademark, it is no less correct that they “use” the trademarks in the common sense meaning of the term “use.” Indeed, that is how the search engines sell keyword ads. For just this reason many courts, before and after Merck, have continued to find use in commerce when trademarks are sold as keywords. See, e.g., Google, Inc. v. American Blind & Wallpaper Factory, 2007 WL 1159950 (N.D. Cal. 2007); J.G. Wentworth, SSC, LP v. Settlement Funding LLC, 2007 WL 30115 (E.D. Pa. 2007); 800-JR Cigar, Inc. v. GoTo.com, Inc., 437 F. Supp. 2d 273 (D. N.J. 2006); Edina Realty, Inc. v. TheMLSonline.com, 80 USPQ2d 1039 (D.Minn.), reconsideration denied, 2006 WL 1314303 (D. Minn. 2006); Gov’t Empl. Ins. Co. v Google, Inc., 330 F. Supp. 2d 700 (E.D. Va. 2004). However, Merck has been followed by other recent cases (primarily if not exclusively in the Second Circuit). See, Site Pro-1, Inc. v. Better Metal, Inc., 2007 WL 1385730 (E.D.N.Y. 2007); Rescuecom Corp. v. Google, Inc., 456 F. Supp. 2d 393 (N.D.N.Y. 2006).

Although the adware companies tend to be a bit more subtle in not expressly selling links to given trademarks, if one excludes those who were born yesterday, it would be hard to believe that the typical business that buys "contextual" pop-up ads is unaware that the contexts in which its ads will likely appear include its competitors' Internet sites. Rather, it is probably a safe bet that most advertisers would not otherwise pay for the service. Just so, on January 30, 2007, New York Attorney General Andrew Cuomo announced a settlement with three advertisers, Cingular Wireless, Priceline.com and Travelocity.com, requiring them, among other things, to pay fines and to exercise due diligence about how their ads are delivered over the Internet. According to Mr. Cuomo, this is the first such enforcement activity against Internet advertisers themselves, rather than simply against the adware firms that place pop-up ads for the advertisers.

One related form of trademark use, comparable to their unseen presence behind keyword and pop-up ads, has been in metatags. Metatags are concealed lines of HTML code that identify the contents of web pages for indexing by search engines. Although typically unseen, metatags can of course be viewed by opening the “file” tab on a browser and viewing the “source” data. When the competitors’ trademarks have been used out of sight as metatags, most courts have not hesitated to hold that such surreptitious use constitutes “use in commerce.” See, e.g., Brookfield Communs., Inc. v. W. Coast Entm't Corp., 174 F.3d 1036 (9th Cir. 1999); Australian Gold, Inc. v. Hatfield, 436 F. 3d 1228 (10th Cir. 2006); Tdata, Inc. v. Aircraft Technical Pub., 411 F. Supp. 2d 901 (S.D. Oh. 2006). But see Site Pro-1, Inc. v. Better Metal, Inc., 2007 WL 1385730 (E.D.N.Y. 2007). Yet, metatags are no more visible to the average Internet user than are the machine linking functions that generate keyword or pop-up ads. It may well be that some uses of metatags or other unseen search engine optimization tools are fair and should be deemed lawful. Indeed, perhaps the clearest example is Bihari v. Gross, 119 F. Supp. 2d 309, (S.D.N.Y. 2000), where the same district court that decided Merck held that unauthorized use of a trademark in metatags was a use in commerce – albeit also a fair use in commerce and hence permissible. In Bihari, the metatags and the site on which they were used employed the trademarks merely for the descriptive purpose of commenting (negatively) on the plaintiff trademark owner. See also Playboy Enterp., Inc. v. Welles, 279 F. 3d 796 (9th Cir. 2002). However, following the Ninth Circuit’s 1999 ruling in Brookfield, in most cases concerning metatags, the unseen use has been deemed to be both “in commerce” and a basis for liability, where there was no fair use defense. Just so, one of the grounds identified in Google, Inc. v. American Blind & Wallpaper Factory, 2007 WL 1159950 (N.D. Cal. 2007) for finding that Google’s AdWords program uses trademarks “in commerce” was the Ninth Circuit’s holding in Brookfield that comparable invisible use of trademarks in metatags is within the reach of the Lanham Act.

Without too badly torturing logic or the English language, concealed commercial use of trademarks in connection with Internet advertising and promotion cannot both be and not be a form of trademark usage – at least not without some significant differences in the actual nature of the uses. Factual differences between the way keyword and pop-up ads are sold have certainly explained some of the differing decisions. Many courts have noted that although adware firms have not offered expressly to link to specific trademarks, such as the firm involved in 1-800 Contacts, the search engines selling keywords typically do. For this and other reasons, most of the courts to have considered keyword ads (including those cited above) have not followed the newly-emerging reasoning of the pop-up precedents; rather, finding such use to be use in commerce for purposes of the Lanham Act (even if unconventional use), such courts have been prepared to weigh liability for trademark infringement.

Because many keyword ads are sufficiently innocuous to present only modest risks of confusing customers; because even those risks that do exist will likely dissipate further as internet users increasingly understand and differentiate keyword ads from native search results; because some such ads are likely fair uses; and because truly misleading advertising should be separately remediable as false advertising, the importance of maintaining analytical clarity regarding what is "use in commerce" perhaps lies less in how keyword ad cases are resolved now than in how such analytical principles will be applied to already foreseeable future cases involving other similarly surreptitious uses of trademarks to guide traffic (and generate revenue) on the Internet. Indeed, as Madison Avenue increasingly comes to rule the Internet (advertising revenues being of course a principal driver of so much Internet commerce), courts will likely come to grapple with ways far more subtle (or perhaps sinister) than keywords or pop-ups that can employ recognized trademarks in unseen ways to help promote competing goods or services. If logical consistency has been somewhat elusive to date, we have yet to see how (or if) courts will regulate such subtler methods, including tools that already exist to optimize the relevancy or prominence of Internet sites as indexed by search engines.

Consider only three such search engine optimization techniques that are already in use: cloaking, invisible text, and link farms. These are sometimes referred to generically as “spamdexing” (which, fortunately, in no way involves attorneys in tight-fitting stretchy clothes). Cloaking is a technical means to present to search engines’ spiders (which crawl the Internet to index content) content different from that presented on users’ browsers. The search engines are effectively tricked into giving the site connected to the cloaked page a higher relevance ranking than its visible content would warrant. And search engine users may also be similarly tricked into visiting a site (for example, a pornographic or infected site) when the search engine description differs – perhaps substantially – from what is shown on the user’s browser. Invisible text similarly may entail unseen uses of protected trademarks to drive Internet traffic. Like unseen metatags, invisible text can be hidden right in the open by presenting words on a website against the same color as the background. Sometimes referred to as a form of “keyword stuffing,” invisible text can cause a search engine to misread the relevance of a page in response to a given Web search and in this way mislead the unsuspecting human audience. Unless the viewer happens to highlight the text with his or her browser, it remains unseen to the human eye, yet it can mislead the colorblind search engines. Finally, because search engines often rank sites based on the number of links to the site, so-called link farms, or groups of web pages that are all hyperlinked to one another, have been similarly developed to increase rankings by unseen manipulation of the coding on a site. Such search engine optimization tools need not employ trademarks any more than metatags, keywords or pop-ups do, but to the extent they all can and do use trademarks in unseen text, how greatly do they differ from one another? Curiously, the search engines that encourage sales of keywords, discourage such spamdexing techniques and sometimes de-list sites guilty of employing such manipulative tools to enhance relevance and rankings.

If the logic of all this seems a little obscure, perhaps it would be best to step back to ponder media other than the Internet. Consider the time shortly after the last great transformation of the information age, when television was new, and a great hue and cry arose over concealed subliminal advertising on the small screen (and in movies). (Consider as well that the book that initially drew such attention to the issue, Vance Packard's The Hidden Persuaders, was written in 1957, as close if not closer to the launch of the TV era than are we to the launch of the Internet era.) The technique typically involved flashing images so quickly the conscious mind could not process them, but the subconscious mind could. In 1974, the Federal Communications Commission defined this technique as any "attempt … to convey information to the viewer by transmitting messages below the threshold level of normal awareness," and concluded that "such techniques are contrary to the public interest." Public Notice Concerning the Broadcast of Information By Means of "Subliminal Perception" Techniques, 44 FCC 2d 1016 (1974). Another example recently in the news (as it was many years ago) involves legal concerns over surreptitious means to influence listening (and purchasing) habits in the broadcast music industry; namely, "payola", the secret "pay-for-play" sums record companies pay broadcasters to promote favored songs. Are either subliminal advertising or payola different in kind from the ever more stealthy techniques employed to drive in Internet traffic?

Keyword ads, pop-ups, and other search engine optimization techniques may all prove to be largely harmless – or at worst, minor annoyances. What is perhaps most troubling about such advertising is the type of software (little different from spyware) used to deliver pop-up ads to the user's computer, and no such software is involved in keyword advertising. Solely as a matter of trademark law, many such uses will undoubtedly be deemed permissible fair uses or appropriate forms of comparative advertising under existing legal rules. Certainly the intellectual property attorneys and other media professionals who devote the most time wrestling with such issues (including those reading this article) are sophisticated enough not to be deceived or misled themselves. Yet, it is less clear how the vast multitudes of Internet users are affected, and virtually none of the reported decisions to date have been based on admissible survey evidence of consumer impressions. As a result, and because of the ever more stealthy techniques being developed to drive Internet traffic, before bright line rules are implemented barring remedies for all such opaque advertising techniques operating below the threshold level of normal awareness, further consideration may be warranted whether use in commerce really does mean something other than commercial use.

The IP Subcommittee co-sponsored a presentation at the 2004 ABA Annual Meeting entitled "Search Engines, Adware and Trademark Law:  What Your Clients Can and Cannot Do in Online Advertising."  The materials from that presentation can be found at http://www.abanet.org/buslaw/mo/premium-cl/programs/ann04/38/38.pdf (ABA membership required).

back to top

In Perfect 10 v. Google, Inc., 416 F. Supp. 2d 828 (C.D. Cal. 2006), the district court granted the plaintiff’s motion for a preliminary injunction against the internet search grant from creating and displaying thumbnail versions of the plaintiff’s copyrighted photos in Google’s image search results. The district court’s decision denying Google’s fair use defense yielded an ironic result: namely, works that were previously available to the public suddenly became unavailable even though Perfect 10’s sale of some of those works was hypothetical. Unsurprisingly, Google appealed the district court decision on several grounds, including the fair use issue. The Ninth Circuit reversed and found that Google will likely be able to sustain its fair use defense because Perfect 10 relied on hypothetical, rather than actual, adverse effects on its market. Perfect 10, Inc. v. Google Inc., Case No. 06-55406 (9th Cir. May 16, 2007); Perfect 10, Inc. v. Amazon.com, Inc., Case No. 06-55405 (9th Cir. May 16, 2007). The case is still in its initial stages, and facts might be developed that could yield a different result under the Ninth Circuit’s decision. Moreover, the Ninth Circuit also determined that the district court applied the wrong test in its analysis of contributory infringement and remanded that claim for further review.

Background

One of many functions performed by Google’s search engine is the delivery of thumb-nail sized images rather than text in response to a query. Perfect 10, an adult entertainment publisher, sued Google and Amazon.com, Inc. (the Court denied the application for injunctive relief as to Amazon because Amazon does not create, store or serve either full-size or thumbnails of Perfect 10’s images, but merely provides a link to the thumbnails stored on Google’s servers) for copyright infringement, alleging that Google retrieved Perfect 10’s copyrighted images and stored and/or displayed them on Google’s server. Perfect 10 objected to Google’s practice because Perfect 10 was exploiting its copyrighted images by selling them for download to mobile phones, where the images appeared in a thumbnail-sized format identical or nearly identical to the version appearing after a search on the Google service. Perfect 10 filed a motion seeking injunctive relief, asking the court to order Google and Amazon to stop displaying thumbnail images of Perfect 10 models in its image search results and to stop users from linking directly to third-party sites that host and serve infringing full-size images.

The District Court’s Decision

The initial matter before the district court was whether Google’s use of Perfect 10’s images was an infringing use. The court began by applying what it referred to as the “server test.” Under this analysis, a search engine such as Google is capable of copyright infringement only if it actually stores the copyrighted images on its own servers. The court determined that Google creates and stores thumbnails of Perfect 10’s copyrighted images on its servers, thus making it liable for copyright infringement. The court applied the server test to the full-size images and determined that Google would likely not be found liable for infringement of the full-size images that are displayed not from Google’s own servers, but via direct linking to third-party Web sites.

Perfect 10 posited a second argument that Google would also likely be held secondarily liable under the doctrines of contributory and vicarious infringement. After a thorough analysis, the district court deemed it unlikely that Google would be found either contributorily or vicariously liable for any use of Perfect 10’s copyrighted images. The district court held that Perfect 10 failed to meet its burden of establishing either that Google would be found to have materially contributed to the direct infringement taking place on infringing third-party sites, or that Google exercised any control over the infringing sites.

Thus, the only remaining infringement claims were related to those thumbnail images that were directly displayed from Google’s servers. And as to those, Google unsuccessfully relied on the fair use defense to justify its conduct to the district court.

The Ninth Circuit’s Decision

On appeal by Google, the Ninth Circuit praised the district court’s handling of a complex case in a “particularly thoughtful and skillful manner.” Perfect 10, Inc. v. Google Inc., Case No. 06-55406 at 1. The Ninth Circuit nonetheless appeared to reassert its holding in Kelly v. Arriba, 336 F. 3d 811 (9th Cir. 2003), and found that the lower court had erred on certain issues. Specifically, the district court erred in concluding that Perfect 10 would meet its burden of demonstrating the likelihood of overcoming Google’s fair use defense, and applied the wrong test to determine whether Google could be held liable for contributory infringement.

Fair Use Analysis

The fair use doctrine 17 U.S.C. § 107 is a defense to a copyright infringement claim that allows the reproduction of copyrighted materials for certain purposes without obtaining permission from, or paying a fee or royalty to, the rights holder. 17 U.S.C. § 107The application of fair use is not absolute; it is contextual and changes based on application. In fact, it has been said that fair use is so flexible as to defy definition. Time, Inc. v. Bernard Geis Assocs., 293 F. Supp. 130, (S.D.N.Y. 1968). The Copyright Act does not define fair use, but rather lists a set of factors to be applied against the objectives of copyright law for the purpose of determining whether the use made of a copyrighted work in any particular case is “fair”: (1) purpose and character of use; (2) nature of copyrighted work; (3) amount of and substantiality of portion used; and (4) effect on the potential market.

The district court found that factors (1), (2), and (4) weighed in favor of Perfect 10, and that factor (3) was neutral. Applying the same fair use analysis to the same set of facts, the Ninth Circuit disagreed, finding that factor (1) favors Google, factor (2) weighs slightly in Perfect 10’s favor, and factors (3) and (4) are neutral.

First, both courts agreed that Google’s display of thumbnails is clearly commercial. Google derives significant commercial benefit in the form of increased user traffic and advertising revenue. However, Google’s use is simultaneously transformative in that it serves the public interest by enhancing Internet searches and the dissemination of other creative works. Upon weighing these competing interests, the district court found that the first factor weighed slightly in Perfect 10’s favor. But the Ninth Circuit reached a contrary conclusion, finding “that the significantly transformative nature of Google’s search engine, particularly in light of its public benefit, outweighs Google’s superseding and commercial uses of thumbnails in this case.” Thus, the Ninth Circuit found that factor (1) favored Google.

Because Perfect 10’s photos are creative, they are subject to greater copyright protection. The images, however, are already published both in print and on the Internet before they appear in a Google image search, making the works more likely to qualify as fair use. Here, the district court and the Ninth Circuit agreed that factor (2) weighed slightly in Perfect 10’s favor.

The two courts agreed upon factor (3), finding it to be neutral, i.e., favoring neither party. Although Google uses an entire image in its application, its use is no greater than necessary to achieve the objective of providing effective image search capabilities.

The greatest divergence between the courts occurred at the fourth factor. Perfect 10 successfully argued to the district court that Google’s thumbnails are essentially the same size and quality as the reduced-size images that Perfect 10 downloads to mobile phones for a fee, and therefore mobile phone users could choose to download free images rather than pay for the same work. On the basis of this argument, the district court held that “Google’s use of thumbnails likely does harm the potential market for the downloading of reduced-sized images onto cell phones.”

However, the Ninth Circuit was not persuaded. It reasoned that because “the district court did not find that any downloads for mobile phone use had taken place,” the “potential harm” to Perfect 10 was “not significant at present.” Thus, the Ninth Circuit found that the fourth factor — effect on the potential market — was neutral insofar as the potential harm remained “hypothetical.”

Upon balancing the significant benefit Google provides to the public against the unproven use of Google’s thumbnails for cell phone downloads, and considering the other fair use factors, all in light of the overall purpose of copyright, the appellate court concluded that Google’s use of Perfect 10’s thumbnails was fair. Finding that Perfect 10 would unlikely be able to overcome Google’s fair use defense, the Ninth Circuit lifted the preliminary injunction.

Contributory Infringement

The Ninth Circuit then reviewed the district court’s finding that Google was not liable for contributory infringement. On this issue, the lower court held that, even assuming Google had actual knowledge of infringing material available on its system, Google did not materially contribute to infringing conduct because it did not undertake any substantial promotional or advertising efforts to encourage visits to infringing Web sites. The Ninth Circuit found fault with this analysis, stating that the lower court had applied the wrong standard. The proper test for determining contributory infringement, according to the appellate court, is the one set forth in Napster: “[I]f a computer system operator learns of specific infringing material available on his system and fails to purge such material from the system, the operator knows of and contributes to direct infringement.” See A&M Records, Inc. v. Napster, Inc. 239 F.3d 1004, (9th Cir. 2001). Applying the Napster test, Google could be held contributorily liable only if: (1) it had knowledge that infringing Perfect 10 images were available using the search engine; (2) it could take simple measures to prevent further damage to Perfect 10’s copyrighted work; and (3) it failed to take such steps.

Because the district court failed to apply the proper test, it did not properly resolve this issue. Thus, the Ninth Circuit remanded this claim to the lower court for further consideration.

Conclusion

After the Ninth Circuit’s decision in Google, it appears that where an infringing use provides a clear and substantial benefit to the public, and the plaintiff cannot show any actual harm, courts should find the use of such works to be “fair.” However, because this case remains at the preliminary pleadings stage, it is conceivable that Perfect 10 might overcome the mere hypothetical harm it alleged in the fourth factor of the fair use analysis and provide evidence of actual downloading of its works from Google’s service to mobile phones. In the event it does so, it is unclear to what extent such evidence might lead to a finding that Google’s use of the works is not fair. In that case, one can certainly expect the question to be put yet again to the Ninth Circuit, at which time it might also address the issue of the in-line linking to third-party sites, which remains an open issue.

back to top

The opinions in the articles contained in this publication are those of the authors only, and not necessarily those of the editors, the Intellectual Property Subcommittee, the Cyberspace Law Committee, or the American Bar Association.