What if an artificial intelligence (AI) tool provides materially wrong or misleading information that causes a user financial loss? Or a chatbot encourages or facilitates self-harm, potentially causing serious harm to a user?
In both cases, the user may have reason to question the service's safety or quality. But not if the AI tool is offered entirely free of charge. This is where India’s consumer-protection laws could present a legal gap.
The Consumer Protection Act, 2019, India’s key consumer-protection law, which provides remedies for defective goods and deficient services, was not drafted specifically with AI in mind. Its definition of “service” excludes services rendered entirely free of charge.
“A person paying for a premium AI chatbot subscription would be availing a service for consideration and could potentially seek remedies under the Act. However, if another person uses the same chatbot entirely free of charge, there may first be a question as to whether they qualify as a ‘consumer’ under the Act," said Ketan Mukhija, partner and co-head private equity and venture capital, at law firm Kochhar & Co.
In simple terms, before examining whether an AI tool has caused harm, it must first be established whether the user qualifies as a “consumer” under the Act. While the issue existed even before AI, it has become more significant as AI tools have grown rapidly in use across the country.
ChatGPT crossed 100 million weekly active users in the country in February 2026, according to OpenAI. It is also Anthropic’s second-largest market globally for Claude.
However, one may argue that AI services are rarely entirely “free”. An older Supreme Court ruling, IMA v. V.P. Shantha, delivered on 13 November 1995, held that a service provided free of charge to some people can still qualify as a “service” if other customers pay for it. This could potentially extend to freemium AI tools, where basic features are free but users pay for additional functionality.
“Most AI services have a paid tier or are bundled with other paid packages, e.g. Copilot inside a paid Microsoft 365 plan. IMA v. V.P. Shantha held that a free service rendered in an establishment where others pay is still a ‘service’. This will cover freemium services,” said Anupam Shukla, partner at law firm Pioneer Legal.
Who to blame?
Another challenge is determining who bears liability. Under the Consumer Protection Act, 2019, consumers can seek compensation for harm caused by defective products or services, with liability potentially resting on manufacturers, sellers and service providers. The question, therefore, is whether this framework adequately covers AI-related harm, given the number of entities involved in developing and deploying AI systems.
Experts say determining who is responsible can be complicated because an AI system can involve several players, from developers and designers to companies that deploy the technology.
“Where the supply itself is an AI-enabled use case such as an application, the liability could be affixed to developers, designers, and deployers, depending on where causation arose, where assessment is likely to move beyond the Consumer Protection Act to cover assessment around responsible AI principles, data protection, IP, and other relevant laws,” said Arya Tripathy, partner at law firm Cyril Amarchand Mangaldas.
However, the question becomes more complicated when AI is used only as part of another product or service. In such cases, holding the AI provider liable could require consumers to meet a higher burden of proof, Tripathy added.
“In terms of proportionately distributing liability, the final service provider deploying the AI system, which faces the consumer, may be held principally liable, although the definition of a product under the Consumer Protection Act leaves room open for upstream liability across the AI lifecycle,” the authors of a paper titled AI and Consumer Rights in India, published by the Centre for Responsible AI at IIT Madras, told Mint.
The ambiguity also extends to how AI systems themselves fit within the Act. The Consumer Protection Act defines a “product” broadly, but its statutory framework is principally structured around goods, products and services in their conventional forms.
“This creates questions regarding the treatment of software and AI systems, particularly because the Act does not expressly define software or AI as a distinct category of product. While certain forms of software may potentially fall within the concepts of ‘goods’ or ‘product’, particularly where software is supplied as a standalone or embedded product, software may also form part of, or be delivered through, a ‘service’ arrangement,” said Mukhija.
How to fight dark patterns?
Another concern is dark patterns. India’s Central Consumer Protection Authority (CCPA) guidelines cover 13 practices that can undermine consumer choice, but experts say they may not fully address AI-driven manipulation. For instance, an AI agent that flatters and steers users towards a product may not fit existing categories such as “confirm shaming”.
“Since the existing guidelines may not capture AI-driven manipulation, the CCPA must eventually recognize such practices. Hyper-personalized and dynamic interactions powered by generative AI could use anthropomorphic features to influence consumers through psychological steering,” said Abhivardhan, founder of Indic Pacific, a technology law consultancy, and president of ISAIL.IN, an AI standards community.
India, however, may not need an entirely new regime.
“The country should strengthen the existing framework with clearer allocation of AI liability, greater transparency and mechanisms to address the evidentiary challenges of AI-related claims. The European Union offers useful lessons. AI may make technology more complex, but it should never make accountability more difficult,” said Mukhija.
The European Artificial Intelligence Act, which came into force on 2 August 2024, is the world’s first comprehensive AI law and regulates AI systems based on the risks they pose to individuals. It applies to developers, deployers and importers of AI systems operating in the EU market.