Artificial Intelligence meets Trademark Law: Unpacking Consumer Autonomy and Keyword Advertising

At the base of this topic lies the agency of a consumer in buying products. For a long time, consumers had autonomy over their purchases. With the boon and the drawbacks of artificial Intelligence, this autonomy is unknowingly circumscribed. Today, our purchasing decisions are inadvertently governed by Artificial Intelligence (‘AI’) due to its pervasive and subliminal nature in a technologically dependent marketplace. Keyword Advertising exemplifies not only the embeddedness of AI in the purchasing process but also where and how trademark law and AI interact. Search engines like Google display websites under a ‘sponsored links’ heading, a feature of their paid referencing service through which advertisers purchase keywords to prioritise their websites over competitors. For example, an advertiser might buy a generic keyword ‘trip’, ensuring their website appears at the top whenever the keyword corresponds to the search terms. However, issues arise when well-known trademarked words, like ‘Polaroid’ or ‘Frappuccino’ are sold as keywords. This can lead to consumer confusion about the source of the products or services, dilution of the brand, etc. Since trademark law, at its heart, deals with how goods and services are purchased, interference in that very process by AI powered Search Engines is bound to raise pertinent questions. Although cases arising from keyword advertising differ factually, they consistently present questions regarding use of the trademark, existence of consumer confusion or deception and most importantly liability remain. This article examines some of the prominent national and international legal precedent to explore how courts have addressed these questions.

International Precedents

The case of Google France, Google Inc. V. Louis Vuitton Malletier1 (‘Google V France’) sheds light on the question of liability of search engines in relation to trademark infringement. Herein, the search engine Google offered keywords for sale in consideration for a fee termed ‘maximum price per click’, determined by the number of times the advertiser’s link is clicked by the internet user. The same keyword could be bought by several advertisers, but the order of the links was decided purely on an economic basis and the quality of the ads as assessed by Google. The Claimant, the renowned luxury brand Louis Vuitton, accused Google of selling keywords linked to its trademarks, as well as terms like ‘imitation’ and ‘copy’, leading to a dispute. The questions before The European Union Court of Justice were -

Does Google’s use of storing the trademarks as keywords for the purpose of selling constitute as ‘use’ within the meaning of Article 5(1) and (2) of Directive 89/104 relating to trade marks or of Article 9(1) of Regulation No 40/94 on the Community trade mark. Further, what is the Google’s liability.

The Court noted that although Google permits the sale of deceptively similar keywords and provides the necessary technical support, this does not constitute 'use'. It is the advertiser that uses the trademark in its commercial communication to the public, since they buy the keywords so that their website comes at the top and the internet user becomes aware of the advertiser’s goods/services in comparison with its competitors. In furtherance of this rationale, the Court held that the purchase by the advertiser and displaying the trademark in its advertisement constitutes use, in relation to the goods and services - the reason of the advertisement, i.e., the advertiser may be seeking visibility amongst its competitors or attempting to pass off its goods/services as that of the proprietor of the trade mark, therefore, there is use 'in relation to goods or services'. Unlike Google, which merely stores the keywords for sale in ordinary course of business, the advertiser is the one using the trademark in relation to its goods/services and thereby causing consumer confusion and an adverse effect on the trademark proprietor’s brand and the trademark’s origin function.

Further, on the question of search engines liability as per Article 14 of Directive 2000/31 of the Directive on electronic commerce, the Court held that the question of liability is governed by whether the search engine has played an active role such that it has knowledge of, or control over the keywords it stores. In other words, is its role merely technical, automatic and passive indicating lack of control over data stored. The Court did not go into determining this question, as it fell under the jurisdictional ambit of the national court, leaving a window open for interpretation on what is meant by having knowledge or control over the keywords stored. The judgment suggests that search engines would be secondary infringers unless neutrality, i.e. the knowledge or control over the infringement, is abandoned.

On the other side of the coin lies the case of Cosmetic Warriors Ltd and Lush Ltd V. Amazon.co.uk Ltd and Amazon EU Sarl2 (‘Lush Ltd V. Amazon’) which compares keyword advertising and unauthorized use of a trademark by an e-commerce website. In this case, the Claimants owned the trademark “Lush” for its bath bombs under Class 3 of cosmetics and toiletries. Amazon sold several products, except all the Claimant’s products, barring hair extensions that were sold through third party sellers. The Claimant alleged infringement arising from three kinds of scenarios. The first scenario involved Amazon buying keywords, including the Claimant’s trademark 'Lush,' such as ‘Lush Soap at Amazon’ or ‘Low prices on Lush Soap’, which appeared on search engine result page. The second scenario was wherein the mark ‘Lush’ as a was not bought or shown by Amazon in its advertisement, but the internet surfer would be directed to Amazon’s website for products similar or identical to Lush products. The third scenario was wherein Amazon‘s website suggested products including ‘lush bath bombs’ and ‘lush hair extensions’ as soon as the keywords were even partially written, despite not selling the Claimant’s products. For example, for any consumer that typed ‘lu-’, a drop-down menu appeared suggesting products under the related searches section and a list of other similar brands that invariably contained the term ‘lush’. Thus, the consumer will expect to see the Claimant’s products as related to the input search term.

Comprehensively, the High Court of England & Wales looked at how the use of the keyword ‘Lush’ affects the origin function of a trademark and the average consumer. With respect to the first scenario, the consumer types the word ‘lush’ as the search term, expecting to be directed to products of the Claimant on Amazon’s website. Further, Amazon did not give any disclaimer as to the fact that it was not selling the Claimant’s products. Thus, this scenario directly leads to consumer confusion and adversely affects the origin function of the Claimant’s trademark.

For the second scenario, it was held that due to absence of the term ‘lush’ on Amazon’s advertisement, there is no consumer confusion since the average consumer does not expect the Claimant’s products on the website.

With respect to the third scenario, a line of difference was drawn between keyword advertising through a search engine and advertising on an e-commerce website. Amazon argued that the Claimant’s attempt to remove its trademark from Amazon’s website is interference in how it decides its search process. The Court, however, held that technological development does not allow a trader to run roughshod over intellectual property rights. Amazon’s software was used to predict consumer search terms, for example, the display of ‘lush’, as soon as the search term was partially written or displayed under related searches. This counts as use of the ‘lush’ trademark in the course of trade and the average consumer would be confused as they expect the products of the Claimant to show on the website when they type in the term.

Therefore, the above two examples show how search engines use algorithms that suggest various results leading to likelihood of consumer confusion and harm to the brand. The standard of likelihood of consumer confusion was kept low assuming that the consumer filling in a particular search term is necessarily looking for the goods/services of a particular proprietor. In the Google V France case, even though there was considerable interference in the purchasing process by Google as it permitted the sale of deceptively similar or identical keywords, further sold them in exchange for a maximum visibility above the actual websites, the search engines escape liability due to their relatively neutral position in using the trademark, as compared to the advertisers. However, more actively engaged platforms like Amazon, in the third kind of infringement, will be held liable as their software directs the consumer towards products they are not looking for, causing confusion. Further, that also leads to brand dilution and affects the origin function of the trademark.

Indian Precedents

The key question that needs to be tackled now is: how can Indian Trademark law hold Artificial Intelligence accountable for trademark infringement? Currently, India lacks specific laws that directly address AI-related infringement. In fact, many AI-driven entities, such as search engines, manage to sidestep liability due to the ‘safe harbor’ provision for intermediaries shielding them from liability. However, there is some legal groundwork that sets a precedent, particularly regarding the degree of involvement in the infringement.

Take, for instance, the landmark case of Shreya Singhal v. Union of India3 , which clarified that intermediary liability is triggered when the intermediary has actual knowledge of the infringement. This is a similar principle as held in the case of Louis Vuitton V. Google France.

In Google LLC v. DRS Logistics (P) Ltd. & Ors. 4 , a more complex situation unfolded. Here, the keywords associated with the Claimant, such as ‘Aggarwal Packers and Movers,’ were being used by third-party websites. These unrelated websites appeared at the top of search results, confusing consumers into believing they were visiting the Claimant's official site. The Claimant therefore alleged that Google, through its Ads programmes, encourages use of its trademarks as keywords by third parties, diverting traffic from its official website to other websites leading to deception of its customer. Google’s defence rested on the argument that since the keywords were invisible to users, they did not meet the legal definition of "use of a mark" under Section 2(2)(b) of the Indian Trade Marks Act, 1999. Moreover, they claimed that the content of the ads is the essential factor in deciding likelihood of confusion, and that any trademark use was actually carried out by advertisers, not by Google. Further, Google being a passive intermediary, is also entitled to the protection from liability offered under Section 79 of the Information Technology Act, 2000. The Delhi High Court, therefore, had to ascertain whether - there is use of the trademark as keywords. If yes, who uses the trademark and whether the use amounts to infringement. Lastly, is Google absolved of its liability.

On the question of ‘use’ as per Section 2(2) of the Trademarks Act, Google contended that its conduct does not constitute as use since the keyword is not visible to the consumer and the section mandates that use must be either visual or printed. The Court negated the argument presented as the term ‘use’ under the section is understood in relation to the goods/services associated with the keywords sold for and its effect on the consumer. As per Sections 2(2)(c)(i), the term ‘use’ is qualified by the terms ‘in relation to’ and ‘goods. Thus, there is use of the mark as a keyword. For who uses the mark, Google argued that there is no use of the trademark by drawing a parallel with Meta-tags. The case of Christian Louboutin Sas V. Nakul Bajaj & Ors.5 , shed light on the workings of meta tags. Herein, meta tags were used instead of keyword referencing services, leading to trademark violations. Meta tags, which are hidden bits of data used by search engines to rank websites, can be invisible to the average internet user unless they delve into the website’s code. But despite being hidden, their impact on consumers is almost indistinguishable from keyword referencing. Thus, merely because the trademark is invisible to the consumer does not conclude that there is no use of the trademark under the Act.

On the question of infringement, the Court reached the conclusion that the use of trademarks as keywords does not per se constitute infringement. The underlying assumption of the Claimant’s argument was that the internet user is looking for the proprietor of a trademark when they feed in the search terms which is why when they see results containing the trademark, there exists likelihood of confusion. The Court disagreed, holding that there could be many reasons, such as reviews on the trademark proprietor or websites offering similar products, etc; Ultimately, the primary role of an internet service provider is to give results based on the relevancy of the search terms, acting as an indexing service for all the website addresses relevant. Use of a trademark as a keyword can also be equated to a real-life situation wherein one product is placed over other similar/identical competitive products in the supermarket, which is not illegal. Additionally, an internet user’s participation in the search result cannot be ignored, as per the Court, anyone surfing the internet knows all search results may not be relevant. A third party cannot extend its rights to the extent that it governs Google, a proprietary internet service provider operating for its own benefit. Thus, as per the rationale advanced above, the keyword is not performing the function of source identification thus, use of the trademark as keywords does not constitute use as trademark under Section 29(1). Use of a trademark as a keyword will lead to infringement only if it causes confusion, dilution, or compromises the trademark,

The Court, therefore, gave consumer autonomy and agency an upper hand. Undeniably, the overt address of consumer autonomy also raises the standard of likelihood of confusion. Most consumers in India may not be internet literate or aware of the algorithms in place, such as browsing history, history, sponsored links, that decide how the search results are displayed on the Search Result Page. In the age of constant technological development and the rise of dark patterns, keyword advertising may become dangerous for consumers and brands alike. There needs to be a higher duty of care in place wherein Google or any other search engine, permits unrestricted use of trademarks as keywords.

This case also addresses that while intermediary platforms like Google can operate within the bounds of the law, their liability hinges on the knowledge and control they exert over potential infringements. Google, in this case, was held to be an active participant in the use of the trademarks by the proprietors as it sells the use of the trademarks and encourages users to include those trademarks as keywords for displaying the ads to the target audience. Therefore, they are not protected under Section 79(1).

Conclusion

As we move forward, it's essential to recognize that the rise of AI technologies, particularly in voice search, could further complicate these issues. For example, if a consumer is asking for a particular brand’s product, there is a possibility that the application suggests an alternative product for a fee paid by a competitor’s proprietor, similar to keyword advertising.

As AI continues to reshape consumer interaction and purchasing processes, nuanced challenges arise such as calibrating likelihood of consumer confusion, brand dilution and liability of search engines in an online interface. Unlike thirty years ago, the market for purchasing goods and services is slowly evolving towards digitization, raising practical issues, that necessitate addressing the effects on trademark law.

Google's AdWords and similar keyword advertising strategies exemplify a reduction in consumer autonomy. The international and Indian case laws discussed, show multiple ways this interaction manifests. In Google France, Google Inc. V. Louis Vuitton Malletier, the Court narrowly interpreted the law, holding that Google’s storage of Vuitton’s trademarks for sale does not constitute as use. Rather, it is the advertiser using the mark in relation to the goods and services, leading to likelihood of confusion. Cosmetic Warriors Ltd and Lush Ltd V. Amazon.co.uk Ltd and Amazon EU Sarl is a step ahead from the former case, herein the Court held that Amazon, as the advertiser, predicts the search terms identical to the Claimant’s trademarks on its website, despite not selling the goods corresponding to the term, leads to consumer confusion and diversion of business from the actual owner of the keywords. However, the Indian Courts have ruled on similar lines of Google France, Google Inc. V. Louis Vuitton Malletier, an example of which is how the Delhi High Court held that Google did not use the trademarks of the Claimant as keywords in the case of Google LLC v. DRS Logistics (P) Ltd. & Ors. The Court, herein, does not account for AI’s ability to use the information we feed as a means to direct or control purchasing decisions. However, on a brighter side, the jurisprudence on intermediary obligations and liability curbs the woes keyword advertising. A balanced regulatory framework is essential to address these challenges, ensuring AI-driven systems uphold trademark rights while minimizing consumer confusion and maintaining fair competition.