INTRODUCTION
The rise of artificial intelligence has rewritten the rulebook on creativity, giving rise to AI generated trademarks that disrupt the core principle underpinning intellectual property protection in India. As Artificial Intelligence actively engages in the field of brand creation, distinguishing between genuine innovation and imitation becomes unclear, raising critical questions about the originality and the lawful ownership of these generated trademarks.
As Indian legislation struggles to keep pace with these quick technological advancements the legal framework faces ongoing uncertainty as well as heightened risk. This piece analyses the transformative effects of AI on trademarks, and provides practical measures to reform and protect India’s intellectual property system for the future.
EXPANDING THE AMBIT OF TRADEMARKS IN INDIA
The landscape of intellectual property in India is going through a noteworthy transformation, mainly in the recognition of non-traditional trademarks. Businesses are making customised marks to appeal to the customer base, laying the way for unconventional trademarks that focus on the themes of touch, smell, vision, and sound. Image trademarks, motion marks, and sound marks are also emerging new types of marks offering new kind of protection to trademarks.
The protection to trademarks is established in the Trademark Act 1999, particularly section 2(1)(zb), which describes a trademark as a whole include any mark consisting of a graphical representation and of differentiating goods or services. Additionally, section 9 sets off absolute grounds for denial, making uniqueness the essential requirement; at the same time, section 11 lays down relative grounds barring marks from getting registered that are close to the earlier ones and can be harmful to the reputation of the previous well-known marks. These provisions have unlocked the door for unconventional trademarks, with the condition that they meet the criteria of uniqueness and non-confusion.
AI, nevertheless, adds an extraordinary twist to this progression, where AI may produce marks that are common due to the retention of existing standards and styles, and the chances of unintentional plagiarism are quite high. Currently, Indian IP laws do not provide a clear framework for trademarking AI-generated materials. As a result, when AI-generated trademarks come close to an already registered trademark, it makes the original registration less feasible and the new mark vulnerable to a suit of infringement.
LEGAL AND ADMINISTRATIVE RECOGNITION OF NON-CONVENTIONAL MARKS
Case laws in India have played a substantial role in revolutionising the trademark practice and procedure over time, for instance, in the case of Yahoo Inc v Akash Arora & Anr, where the Delhi HC gave a landmark judgement establishing for the first time that a domain name is entitled to the same degree of protection as a trademark, also in the case of Tata Sons Ltd. v Manoj Dodia & Ors the Delhi HC once again in the year 2011 empahsised the significance of phonetic similarity, highlighting the need for AI systems to accurately assess sound based confusion.
Further the SC in the case of Cadila Healthcare Ltd. v Cadila Pharmaceuticals Ltd laid down key principles regarding deceptive similarity and public confusion establishing a strong base for AI algorithms assessing likelihood of confusion, also the case of Satyam Infoway ltd v Siffynet Solutions ltd the SC court acknowledged that trademarks are not limited to goods but they cover new technological identifiers, and the court reiterated that nature of protection is to protect the aspect of distinctiveness and prevent deception, which are the core principles in the age of AI for creating unconventional marks, with that early developments have also been seen in recognising non-traditional like sounds, smells, shapes, motion, etc., making them registrable signs.
The registry itself has modernised itself by making the majority of the process online, through which 95% IP filings are done in recent times. Also, the government has introduced AI and Machine Learning (ML) based Trademark Search Technology for the faster disposal of applications. In 2024, India saw a rise of 9.3% in trademark applications from the previous year, with trademarks of sound, colours, and animations becoming more popular. Globally, countries such as Australia have already implemented AI-driven tools, including the Trademark International Classification Service (TMICS) and Trademark Precedent Identification (TMPI), in their systems, offering far quicker acceptance for unconventional marks.
Whereas and member countries of the European Union (EU) has implemented AI tools like AI image Search tools and TM class for automated classification, reducing the errors in classification by 40%. Compared to these, India’s approach has been gradual and requires careful navigation and thoughtful harmonisation of legal, ethical, and governance reforms in the system. Reiterating the point that India does not have a clear context concerning AI-generated trademarks leaving the question of registration, ownership, and enforcement in doubt.
THE UNIQUENESS CHALLENGE IN THE AGE OF AI
Trademarks in India must be unique and non-obvious to enjoy protection. Yet, AI-powered design tools are trained on vast datasets that can create visually appealing logos by considering design principles, colour theory, and current trends, which can resemble already existing works. Models are trained on massive corpora of data, much of which is copyrighted material scraped from the internet. AI systems can rapidly generate a large number of trademarks; however, due to their dependence on existing data and a lack of contextual or nuanced understanding, they can create similar marks, which increases the risk of infringement because of the sheer volume of output and algorithmic limitations.
This phenomenon can be seen as the cause of why numerous AI systems might produce over-lapping or closely matching marks. These patterns make the work of the court somewhat difficult, as they find it hard to apply the average consumer test, but through judicial precedents, they have framed some factors that will be seen before applying the test, which include the nature of the marks, their degree of resemblance, class of purchasers, etc.
But these concepts, like the average consumer test, are still not enough to cover the artificial consumers; the courts must redefine these outdated concepts to incorporate artificially generated consumers. Without such adoption or incorporation, the speed and volume at which these marks are being generated have the potential to threaten the market, and it can lead to the ultimate dilution of brands that are authentic and legitimate.
CLARIFYING AUTHORSHIP AND OWNERSHIP CLASHES IN AI-GENERATED TRADEMARKS
Indian trademark law, under the Trademarks Act, 1999, inherently requires ownership by a person who is defined as an individual, a company, a partnership, or other legal entity. AI systems, lacking such legal personality, are therefore not eligible to hold ownership rights over trademarks themselves. This statutory gap raises the critical question: Who can be the owner? Can it be the developer, user, or commissioning party? For most businesses and individuals, it’s either the user or the commissioning party that is considered to be the owner, provided they can prove some human intervention in the approval of the trademark. The relevance of this debate is illustrated in the dispute over Flaxxi AI’s opposition to OpenAI’s trademark application for Chat GPT, where the startup asserted prior use of the name since 2022 in the education sector.
These clashes underline and complicate the traditional notions of authorship and ownership, specifically when the human intervention is partial to secondary or negligible creative input is there from the side of humans, raising questions about who lawfully holds the trademark when AI plays the main part in the creation of a mark.
The ideal practice in the industry states that when a human’s input (prompt) demonstrates original creativity and control over the final output, it is most likely that the courts will assess the matter, and there are high chances that the human will be given the authorship or the ownership of the work. Virtually, this means that the human has to have evidentiary proof, like inspired briefs, evidence of human editing, to establish their claim of authorship or ownership.
This mix model, combining AI help with provable human writing, imitates international best practices. It guarantees that while novelty prospers, the trustworthiness of the trademark system is well-preserved against the influx of authorless AI-generated marks.
CASE STUDY: ANI V OPEN AI OWNERSHIP DISPUTES IN THE AI ERA
As discussed earlier, AI-generated marks raise fundamental questions about authorship and ownership under Indian trademark law. The case of Asian News International (ANI) v OpenAI has brought these issues into sharp focus. ANI alleges that the storage of data, which is in the form of news by OpenAI to train its Large Language Model (LLM) model ChatGPT to generate responses, is an infringement of their copyright. While the primary dispute concerns copyright infringement, the implications for trademarks are noteworthy: if an AI uses a brand’s distinctiveness fundamentals in its responses, that brand’s goodwill and uniqueness could be damaged.
This case also reflects how AI disputes are escalating beyond copyright and patents, touching upon the core of trademark protection. The Delhi High Court’s discussion in this matter has included whether accountability can be placed on AI developers, deploying companies, or users commissioning AI responses. Although a final ruling is pending, the case emphasises the pressing need for statutory lucidity on attribution and accountability in AI-assisted marking.
GLOBAL APPROACHES TO AI GENERATED TRADEMARKS
Internationally, the regulation of AI generated trademarks is still evolving, but diverse approaches offer useful lessons for India. In Japan, the patent office has started an AI-based prior image trademark search system that has been in use since April 2023, but the same has not been extremely successful, data shows that trademark application filings declined following the JPO’s April 2023 introduction of its AI based image search tool: 170,275 applications in 2022, 164,061 in 2023, and 158,792 in 2024,.
However it was a minimal decrease but yet it was noticeable. The European Union Intellectual Property Office has also started using AI tools like image search tools, chatbots that can help in the E-filing of applications. Figures show a marginal rise in the number of filings since the introduction of these new features. The EUIPO, office received 180,451 EUTM applications in 2024, about +2.7% versus 2023.
This shows that the integration of AI in trademark might not be that successful in Japan but was in the EU. Yet, if such integrations related to trademark is done in the Indian Trademark System the results would be very different, with a high probability that there will be a surge in the number of trademark applications in the country, however for this to happen all the stakeholders need to work and get some crucial reforms in the primary legislature which is governing the said law in the country.
CLOSING THE POLICY GAP PATHWAYS FOR REFORM
One such reform opportunity would be to amend Section 18 of the Trade Marks Act to require applicants to reveal the role of AI in mark creation, permitting inspectors to evaluate human creative input, another reform can be done in section 9 and 11 of the trademarks act where the legislature under section 9 can introduce a new sub-section where they can bring a AI Distinctiveness Index (AIDI), this index would measure a mark in respect to its intrinsic uniqueness and not merely asses on the similarity of the mark. Likewise, in section 11 there can be an Automated Conflict resolution protocol (ACRP) that can be brought which will pre-emptively detect possible conflicts between an already existing mark and the AI generated mark.
But these reforms can take a longer than expected, if we talk about smaller steps we can start by adding a column in the new TM-A form where the person can disclose the use of AI before sending the application for review. These reforms if implemented properly and efficiently can significantly improve the trademark search and registry process in the country.
CONCLUSION
The Indian Trademark sphere is experiencing significant alteration in the sense that it is adjusting to new types of mark, especially the unconventional types. In this context, trademarks created with the help of AI introduce unique challenges especially where there is no specific statutory outline to govern the same. Since AI thrives on the use of large data sets, it can render identical marks making the average consumer test not accurate. The possible solutions that can pave the way for a more robust check mechanism for AI generated trademarks can be amending the trademarks act to include a provision of AI revelation and introducing new standards of assessing the uniqueness of a mark.
Author Details: Devansh Bansal and Sanidhya Ranjan, 3rd Year Students, Symbiosis Law School, Nagpur.
Read Our Latest Article: JUDICIAL POWERS AND MODIFICATION OF ARBITRAL AWARDS UNDER THE ARBITRATION ACT: GAYATRI BALASAMY V. ISG NOVASOFT TECHNOLOGIES LTD. 2025 – J.P. Associates