AI-Generated Trademarks: From Prompt to Protection
- Aaina Sethi & Divanshi Gupta
- Jul 18
- 5 min read
INTRODUCTION
The conceptualization and creation of a trademark has long been anchored in human creativity. Conventionally, branding was an intentional and collaborative process. Businesses would work closely with designers and strategists to craft identities that were distinctive, memorable, and aligned with their long-term vision. A trademark was more than just a logo or a name; it represented a journey of thoughtful ideation, strategic clarity and clear authorship. The traditional landscape is now undergoing a marked transformation with the emergence of Generative Artificial Intelligence (AI) Platforms such as Midjourney, ChatGPT, DALL·E, Adobe Firefly, Canva, etc. have made it possible to develop a brand identity with minimal input, sometimes requiring only a few prompts, or even a single one. However, this evolution brings with it a host of legal complexities.
DISTINCTIVENESS AND THE CHALLENGE OF ALGORITHMIC GENERICISM
AI driven tools function by analyzing extensive datasets to generate brand elements based on user inputs. While this allows for efficient content creation, it also significantly increases the chances of producing results that are similar to existing trademarks or to those generated for other users with comparable prompts. For instance, an AI tool prompted to create a name for a skincare brand may generate results such as Glowsense, Dermaglow, or Skinique, which closely resemble existing marks or follow familiar naming trends within the industry. This raises serious concerns regarding the distinctiveness of AI generated trademarks. The tendency of these systems to rely on common patterns and overused structures results in what is referred to as ‘algorithmic genericism’. Such outcomes are often indistinctive, generic, and repetitive, making it difficult for the resulting marks to serve as effective source identifiers and increasing the likelihood of consumer confusion. These concerns highlight the growing difficulty in securing and protecting AI generated trademarks, especially when distinctiveness is compromised by the repetitive and data-driven nature of such outputs.
CHALLENEGES FACED BY GLOBAL TECH COMPANIES IN SECURING TRADEMARK PROTECTION IN INDIA
Global AI giants such as Google, OpenAI, etc. are encountering difficulties in securing trademark protection for their models and chatbots in India. For instance, Google applied for registration of the mark “GEMINI” under Application Number 6323275 in Class 09. Upon examination, the Trade Mark Office raised an objection citing the prior "Gemini" marks of Sun TV Network Limited, which operates in the television broadcasting sector. In response, Google argued that its AI software and Sun TV’s broadcasting services fall within entirely different industries, with no overlap in customers or trade channels. Similarly, OpenAI’s application for the mark “CHATGPT” under Application Number 5870117 was opposed by FLAXXI AI Private Limited, an Indian startup claiming to be the prior adopter and user of the mark “CHATGPT.” These developments reflect the complexities even leading global players face in protecting their core AI model names. Given these developments, it is evident that AI generated content and trademarks may also face higher scrutiny and similar challenges in the Indian registration process.
DETERMINING OWNERSHIP OF AI-GENERATED TRADEMARKS
One of the central legal challenges associated with AI-generated trademarks lies in determining ownership. While most generative AI platforms include terms of use that either assign rights to the user or grant a license to the generated output, such provisions do not independently establish ownership under trademark law. For instance, OpenAI’s terms for ChatGPT specify that users retain ownership of the outputs.[1] However, such private contractual arrangements do not independently establish trademark ownership under statutory regimes.
Under the Indian Trade Marks Act, 1999, ownership of a trademark can be acquired by any “person,” which includes individuals, companies, firms, and other legal entities. The Act does not address the mode of creation of the trademark. In practice, ownership is recognised in favour of a legal entity that applies for the mark and uses it in the course of trade, irrespective of whether the mark was created manually or with the assistance of AI. Similar principles apply under the frameworks followed by the United States Patent and Trademark Office (USPTO) and the European Union Intellectual Property Office (EUIPO). In both jurisdictions, the method of creation is not determinative, but ownership must vest in a legal person capable of holding rights and assuming responsibility for the mark's use and enforcement.
RISKS OF PASSING OFF AND INFRINGEMENT
AI generated trademarks also raise significant concerns related to infringement and passing off. Generative tools function by analysing large datasets that often include existing brand names, logos, and design elements. As a result, they may unintentionally produce outputs that closely resemble existing registered or well-known marks. Such similarities, even if unintended, can lead to legal disputes, particularly in cases where there is a likelihood of consumer confusion. One such instance involves The Walt Disney Company[2], which voiced concerns about potential trademark infringement arising from AI generated content. In that case, social media users were engaging with Microsoft’s Bing Image Creator, powered by DALL·E 3, to generate illustrations of their pets in a style reminiscent of Pixar animations. Disney reportedly took issue with instances where the AI inadvertently inserted the Disney-Pixar logo into the generated images. This posed a risk of infringing its registered trademarks and underscored the growing challenges of protecting brand identity in the era of generative AI. In India, judicial precedent in this area is still in its formative stages, and further clarity is expected as relevant cases come before the courts.
STRATEGIES TO ENHANCE REGISTRABILITY AND ENFORCEABILITY OF AI-GENERATED TRADEMARKS
Given concerns around lack of distinctiveness and potential overlap with existing marks, it is advisable to incorporate some level of human input when working with AI generated trademarks. Manual refinement of the AI output, whether through conceptual changes or visual adjustments, can help create a mark that reflects intention and originality. This demonstrates that the trademark is not merely a product of automated processes but involves creative judgment that sets it apart from generic outputs. In addition, a comprehensive trademark search remains an essential step before adopting any AI assisted brand identity. Since AI tools often draw from common branding patterns, the risk of similarity with existing marks is elevated. Identifying such overlaps at an early stage can help avoid objections and reduce the likelihood of future disputes. Maintaining detailed documentation of the development process is equally important. Records of prompts used, multiple outputs generated, and any subsequent human modification can be relevant in resolving questions of authorship or good faith. It is also critical to assess the terms and conditions of the AI platform used in the creation of the mark. While some platforms confer full ownership of outputs to the user, others may include limitations on commercial use or retain certain rights over the generated content. Ensuring that the user has complete and unambiguous rights over the trademark is vital for both registration and enforcement. When applied together, these measures offer a structured approach to navigating the legal complexities surrounding AI generated trademarks and enhance their potential for protection in practice. In India, given the growing reliance on AI, it is conceivable that Trade Mark Examiners may begin using AI detection tools such as GPTZero, Scribbr, or similar platforms to assist in assessing distinctiveness and other relevant parameters during examination.
CONCLUSION
The registrability of a trademark depends on its distinctiveness and its ability to indicate the source of goods or services, regardless of whether it is created by a human or through generative AI. However, the growing use of AI tools brings with it the risk of algorithmic genericism, where outputs may fall short of the distinctiveness required for protection. To mitigate this, it is advisable to frame prompts thoughtfully and retain clear documentation of the generation process, which may support claims of prior generation or adoption in the event of an objection or opposition. In the absence of specific legislative or judicial guidance, AI-generated trademarks continue to occupy a legally uncertain space, requiring a cautious and well-informed approach by the Applicants.
References:
Comments