Christian Louboutin SAS & Anr. v. M/s The Shoe Boutique - Shutiq (CS(COMM) 583/2023)

Christian Louboutin SAS and Clermon ET Associes, the French entities behind the Louboutin brand (a trust holding the intellectual property and its exclusive licensee), sued M/s The Shoe Boutique (trading as Shutiq), a partnership firm selling shoes in Delhi malls, in the Commercial Division of the Delhi High Court. They alleged that the defendant had imitated the brand's RED SOLE mark and its "SPIKED SHOE STYLE" designs, and had used photographs of Bollywood celebrities on social media. The plaintiffs said their sales in India in 2022 were approximately Rs 22 crore and that the RED SOLE had already been declared a well-known mark under Section 11(6) of the Trade Marks Act, 1999, in Christian Louboutin SAS v. Pawan Kumar, an order of 12 December 2017.¹

The matter was listed for the first time before Justice Prathiba M. Singh on 22 August 2023, in hybrid mode, on the plaintiffs' application for an urgent ex parte injunction. The defendant appeared through counsel, and one of its partners, Kunal Makkar, was present in court.¹

Counsel for the plaintiffs argued that the reputation the plaintiffs had earned could "also be evaluated on the basis of a chat GPT query which was put on behalf of the Plaintiffs". The query and the chatbot's answer are reproduced in the judgment as images, so their wording is not available in the text version of the order.¹

The defendant then gave a statement through Mr Makkar that the disputed designs had been supplied on a "made to order" basis when customers asked for them, and gave an undertaking, on behalf of the firm and his family members, not to imitate, copy, manufacture or sell shoes imitative of the plaintiffs' designs. The court recorded that it was not recognising a monopoly in all spiked shoes or coloured soles, because an injunction required the impugned products to be "a colourable or a slavish imitation" of the plaintiffs' get-up and designs. In that context the court noted that "when two different questions are put on CHAT GPT, the following answers emerged", again reproduced as images.¹

In paragraph 28 the court drew a general lesson from the three ChatGPT outputs before it, the plaintiffs' and its own two. It held that the responses show that the tool "cannot be the basis of adjudication of legal or factual issues in a court of law."¹

The court gave four reasons. First, an LLM-based chatbot's response depends on the nature and structure of the user's query and on the training data. Second, such chatbots can generate "incorrect responses, fictional case laws, imaginative data". Third, the "accuracy and reliability of AI generated data is still in the grey area". Fourth, "at the present stage of technological development, AI cannot substitute either the human intelligence or the humane element in the adjudicatory process". The court left room for limited use: "At best the tool could be utilised for a preliminary understanding or for preliminary research and nothing more."¹

The court then compared the parties' shoes and a chart of designs, found "a clear intention to imitate and gain monetarily on the strength of the reputation and goodwill of the Plaintiffs", described the defendant's products as "knock-offs or look-alikes", and held the acts to be an attempt to pass off the defendant's goods as the plaintiffs'. It decreed the suit in terms of paragraph 65(a) and (b) of the plaint, held the defendant to its undertaking with Rs 25 lakh payable as damages on proof of a breach, awarded Rs 2 lakh costs because of the defendant's use of celebrity photographs and sale in high-end malls, and ordered a 50% refund of court fees.¹

The reasoning rests on two distinct concerns. The first is evidentiary reliability: a chatbot's answer to a lawyer's prompt is not a verifiable source for reputation or goodwill, which are facts proved by sales figures, advertising, registrations and other records. The court's own experiment, in which two different questions produced different answers, was used to show how outputs vary with the prompt. The second is institutional: adjudication is a human function, and the court expressly refused to let AI stand in for the judge's own reasoning.¹

The passage is limited in several ways. It is obiter dictum, because the suit was decreed on the undertaking and on the court's comparison of the shoes, and the ChatGPT outputs did not decide anything. It does not prohibit the use of AI by counsel or by courts, and it expressly allows use for "preliminary understanding" and "preliminary research". And it is a single judge's observation in an interim-stage order, not a Division Bench or Supreme Court ruling, although the Supreme Court in 2026 took a consistent position when it demanded "a human in the loop at every stage" in Pooja Ramesh Singh v. Jammu and Kashmir Bank Ltd.⁶

The order was delivered five months after the Punjab and Haryana High Court published a ChatGPT answer in a bail order,³⁴ and it set the counter-position early: general-purpose chatbots are not authorities, their outputs are prompt-dependent, and judges must keep the decision for themselves. An SCC Online article lists it among the judicial statements of the principle that AI cannot replace human intelligence in adjudication.²

The observation about "fictional case laws" anticipated a problem that later surfaced in orders of tribunals and courts, beginning with an Income Tax Appellate Tribunal order of December 2024 that cited authorities which could not be found.⁷ By 2026 the Supreme Court was treating reliance on fabricated precedents as a threat to the integrity of adjudication.⁶ Justice Singh has continued to engage with the subject. In November 2024 she was reported as saying at an IBA symposium that AI can be used as a tool by the justice system but that adjudication cannot be outsourced.⁵ On 30 April 2026 a Delhi High Court Division Bench of Justice Singh and Justice Madhu Jain stayed a district court decree against Akasa Air after the airline said the judgment attributed non-existent propositions to Supreme Court decisions, recording that the manner of drafting, "prima facie, gives an impression that there is some AI softwares which may have been used" but that this could not be concluded without further examination.⁸

The suit was decreed on 22 August 2023, and the sources reviewed did not show any appeal. The defendant's undertaking continues to bind it, with Rs 25 lakh payable as damages on proof of breach. The ChatGPT paragraph has not, in the sources reviewed, been overruled or doubted, and later commentary such as SCC Online's April 2026 article cites it for the rule that AI output is, at most, a pointer for preliminary research. As of 15 October 2026, the Supreme Court's draft Regulations for Use of Artificial Intelligence in Courts, 2026 take a compatible line: AI may support research and drafting, but judges alone determine questions of law, fact and justice.