Recently, a DHC Division Bench allowed a Moradabad-based restaurant to continue using Gulshan-e-Karim, subject to a mandatory disclaimer stating that the establishment had no connection with New Delhi’s Karims restaurant chain. Analysing the Court’s findings on what it labelled “innocent infringement,” initial interest confusion, and the anti-dissection rule, Vikram Nanda explains what the judgment has to offer for the Indian trademark regime. Vikram Raj Nanda is a third year student at National Law School of India University, Bengaluru with a keen interest in IP law, Competition Law, and Arbitration. His previous posts can be accessed here.
Garden of God or Garden of Confusion? Analysing the DHC’s Karim Decision
By Vikram Raj Nanda
In an order dated 6th November, 2025, a Division Bench of the Delhi High Court delivered a very interesting judgement in the dispute between the famous ‘Karims’ restaurant chain and Moradabad-based restaurant operating under the name ‘Gulshan-e-Karim’ (translating roughly to ‘Garden of God’). The Commercial Court had earlier granted an injunction restraining the defendant from using the mark Gulshan-e-Karim, holding that the phonetic similarity with KARIM was sufficient to establish a likelihood of confusion.
While the Division Bench upheld some of their findings on infringement and the likelihood of confusion, they ended up modifying the remedy granted. Instead of a complete restraint on the defendant’s use of the word ‘Karim’, the DHC permitted the defendant to continue using its Gulshan-e-Karim mark subject to a mandatory disclaimer stating that their establishment had no connection with the Karims chain. Their underlying reasoning raises a few interesting questions about how the Court applies the doctrine of initial interest confusion (IIC), its handling of the dominant-part versus anti-dissection debate, and its introduction of the term ‘innocent infringement’ when tailoring relief – all of which I examine in this post.
A Brief History
To give a brief background, Karims is a popular restaurant chain that claims to have been founded back in 1913 by one Haji Karimuddin who claimed ancestry to royal cooks during the Mughal era. It has been the proprietor of the mark ‘Karim’ under the Trade Marks Act, 1999 and has amassed significant goodwill and reputation in the market.
The defendant operates their restaurant in Moradabad. Before the Commercial Court, it claimed to have adopted the mark as far back as 1997, though the physical outlet itself began operations only in 2016, and the consistent use of their mark Gulshan-e-Karim appears to date from then. The plaintiff’s chain claims to have learned of this in 2020, after which it initiated infringement proceedings.
The Commercial Court noted the similarities in the impugned marks, noting that the dominant part of the mark was ‘Karim’. Consequently, given that both the entities operated restaurants and had a similar consumer base, likelihood of confusion was apparent and hence, an injunction was granted in favour of Karims.
The DHC affirmed the likelihood of confusion. However, the reasoning adopted by the Court becomes quite interesting. To put it briefly before a more detailed discussion, the Court decided on the following grounds: first, the Court applied the initial interest confusion test (IIC) to hold that there is a case of infringement [I]; second, the Court noted that the well-settled ‘anti-dissection’ principle of trademarks is made subject to the ‘dominant part’ test, as a result of which the dominant part of the marks, i.e.., ‘Karim’ is common in both and likely to lead to consumer confusion [II]. However, and perhaps forming the most interesting part of their decision, the court decided that this was a case of ‘innocent infringement’ and consequently, modified the injunctive relief by the lower courts to instead allow the defendants to use the mark ‘Gulshan-e-Karim’ subject to the clear disclaimer that they had no association with the Karims restaurant chain [III]. Having given a broad overview, let us dive deeper into the case.
Initial Interest Confusion: Going back to thinking consumers are not smart?
The IIC is often used by the Courts to see whether an unassuming consumer will be momentarily confused by the defendant’s mark so as to make an association with the plaintiff’s trademark, even though such confusion may be dispelled later on before purchasing. This fleeting diversion of attention, if demonstrable, is treated as sufficient to constitute infringement (see here).
What becomes crucial, however, is how courts conceptualise the relevant class of consumers and the level of attention they are likely to display (see here). This has been the centre of a long-running debate (see two earlier posts by Aditya Bhargava here and here, summing up the debates very well). Broadly, courts may adopt one of two approaches. Either the courts can treat the modern-day consumer as well-informed, with access to the internet who would ensure that if two marks are similar, which one is the authentic one or what they want to purchase. For instance, the DHC in Under Armour vs Anish Aggarwal (2024)seemed to indicate that consumers today are more sophisticated, and choose even if ‘faced with transient wonderment, to find out what the differences could be’. A similar interpretation had been given in the Mountain Valley Springs India Private Limited v. Baby Forest Ayurveda Private Limited & Ors regarding the sophistication of consumers (though it had later been overturned in a laconic order by the DB).
However, in the Karims case, it appears the DHC moves away from this understanding and adopt a different approach. Its reasoning suggests that it believes consumers may possess a lower level of attentiveness. On the facts of the case, this may not be necessarily incorrect: an unassuming customer walking in search of food may not pause to verify whether the outlet is affiliated with the Karims chain, and may well be momentarily deceived about such a connection. But this very conclusion requires a more structured analysis. As the Under Armour case indicates, courts must delineate specific parameters when assessing the relevant consumer’s behaviour and must not deal in abstractions. The courts should specify who the consumer is and what factors such a consumer is likely to consider.
For instance, one such factor highlighted in Under Armour is the price point of the goods. Do the two establishments in question operate in the same price segment? Did the cost of dining at Karims differ from Gulshan-e-Karim? Was confusion realistically possible in this market context? Some of these factors could’ve been better elucidated by the court to reach its conclusion. Since the Court was sitting appeal, it perhaps chose not to engage deeply with such factual inquiries.
Dominant part vs Anti-Dissection
Another factor that weighed the Court’s conclusion on consumer confusion was the dominance of ‘Karim’ in both the impugned marks. The Court holds that the well-settled doctrine of anti-dissection (as laid down in Kaviraj Pandit Durga Dutt Sharma vs Navaratna Pharmaceutical) is ‘subject’ to the dominant part test. Explained simply, the anti-dissection rule mandates that marks in question be analysed as composite wholes, and mere similarity in some elements will not constitute infringement if the two marks appear different on the whole.
However, in recent times, several courts have indicated that if a certain part of the mark is dominant or more distinctive and likely to influence consumer perception, then the court may place more emphasis on this though it still required to conduct a holistic analysis (see here). These tests are not mutually exclusive and largely, identifying the dominant mark is not solely determinative and remains only an aid for a more holistic analysis. The court is still required to base its analysis on what the marks look like on the whole. The Supreme Court in Pernod Ricard vs KS Chabra, citing McCarthy in Trademarks and Unfair Competition, also echoes this rule, noting that given inherent inaccuracies in consumer perception and recollection, the anti-dissection rule is essential to ensure that courts do not artificially break up composite marks (p.32.2).
Consequently, the DHC’s strong language in this case by noting that anti-dissection is ‘subject to’ dominant part test and mere similarities in ‘dominant part’ may constitute infringement does not sit comfortably with the prevailing law. It suggests that the dominant-part approach operates as an exception to anti-dissection, when in fact it is meant to function only as an interpretive aid.
Furthermore, in undertaking this analysis, the court notes that Karim is the dominant part in Gulshan-e-Karim. The only reason provided is that the word ‘God’ in garden of God cannot be less important than ‘garden’. Notwithstanding this flimsy reasoning here, another important aspect is that courts treat the phonetic similarity due to the presence of ‘Karim’ in both these marks as sufficient for constituting consumer confusion (p.12.7). There is seemingly no discussion on other features of the two marks, including their visual presentation. At the very least, perhaps we can understand this as an instance of a dominant phonetic similarity as being sufficient to constitute infringement, though perhaps other factors were missing.
Innocent Infringement or Merely Injunctive Relief?
Lastly, and perhaps most interestingly, the Court characterises the case as one of ‘innocent infringement’. Finding no indication of intentional, or ‘contumacious’ conduct, it modifies the injunction to permit continued use of the Gulshan-e-Karim mark, subject to the use of a clear disclaimer stating that the outlet has no affiliation with the Karims restaurant chain.
This raises an immediate question: what, in fact, is innocent infringement? The terminology is troubling. Justice Hari Shankar himself in an earlier case (see here) cautioned that infringement under the Trade Marks Act is a statutory tort with no distinction between deliberate and innocent infringement (p. 45.18). Furthermore, the closest analogue to something like ‘innocent’ infringement is perhaps the defence of ‘honest and concurrent use’ (see earlier post here). However, that has its own statutory regime under sections 12, 9(1), 30(1), and 35 of the TM Act, 1999. Yet none of these provisions are discussed in the present judgment.
So the question really is: what was such terminology supposed to indicate? I believe the term innocent infringement is more of a loose descriptor than any new principled category. It appears the Court simply wished to tailor the remedy to the equities of the situation, balancing the plaintiff’s rights with the defendant’s localised, good-faith use. This is hardly unprecedented. In Modern Foods v. Modern Snacks (2023), the DHC modified injunctive relief to the equities of the situation, allowing conditional use of the infringing mark by the defendant. The same instinct perhaps underlies this decision as well, and it is hoped that there is no new category of ‘innocent’ infringement being conjured up here, despite the ambiguous terminology.
In sum, a lot of interesting questions were raised in this order. Perhaps a future court will confront these tensions more explicitly rather than leaving them at the margins of the reasoning.