AI-Generated Content and Copyright Law in India: Can You Copyright What an AI Creates?

Generative AI tools now draft blog posts, design logo concepts, generate product images, and even write code — often in seconds, from a single prompt. For businesses and individual creators in India, this raises two questions that barely existed a few years ago: does copyright protect any of this output? And if it does, who actually owns it — the person who typed the prompt, the company that built the AI tool, or no one at all?

These are not academic questions. If your marketing team publishes AI-drafted blog content, if your startup uses an AI tool to generate a logo concept or app icon, or if your developers lean on AI code-completion tools, you are producing commercially valuable material whose legal status is genuinely unresolved. This matters because copyright is what lets you stop a competitor from lifting your content wholesale — and if the content was never protected to begin with, there may be nothing to enforce.

India's Copyright Act, 1957, was not written with generative AI in mind. But unlike many countries currently legislating from scratch on this issue, India already has one statutory provision — dealing with "computer-generated works" — that predates the current wave of generative AI by decades and is now, somewhat unexpectedly, the closest thing Indian law has to a direct answer. It does not settle the question. But it gives us somewhere concrete to start.

The relevant hook lives in Section 2(d) of the Copyright Act, 1957, which defines who qualifies as the "author" for different categories of work. Section 2(d)(vi) states that for "any literary, dramatic, musical or artistic work which is computer-generated," the author is "the person who causes the work to be created."

This clause was added to Indian copyright law well before tools like ChatGPT, Midjourney, or Gemini existed. It was originally aimed at earlier forms of computer-assisted output rather than large language models or diffusion-based image generators, and no Indian court has yet applied it squarely to a modern generative-AI dispute in a way that gives businesses a clean, judicially tested answer.

Even so, it is the only place in Indian statute that contemplates authorship of a work where a human did not directly write or draw every element by hand. Read plainly, it points toward the person who "causes" the AI output to be created — plausibly the individual who designed the prompt, selected among multiple generated outputs, directed revisions, or otherwise shaped what the tool produced. But "causes to be created" is a phrase that has not been judicially tested in the AI context, and reasonable lawyers can genuinely disagree about how far it stretches. Treat it as a starting point for analysis, not a guaranteed answer to who owns your AI-assisted content.

Even if Section 2(d)(vi) tells us who might be treated as an "author," it does not by itself answer whether the output qualifies for copyright at all. Copyright only protects works that are "original," and Indian courts have been fairly clear about what originality actually requires.

The Supreme Court's decision in Eastern Book Company v. D.B. Modak (2008) is the leading authority here. It rejected the older, low bar of "sweat of the brow" — the idea that mere labour or effort in producing something is enough to earn copyright. Instead, the Court held that a work must reflect a modicum of creativity, involving some degree of skill, judgment, and individual capacity in its creation and arrangement, and not merely mechanical or routine effort.

This matters enormously for AI-generated content. A prompt that simply asks a tool to "write a 500-word blog on GST registration" and publishes the first output verbatim involves very little human skill or judgment in the actual expression that results — the creative choices about wording, structure, and phrasing are made by the model, not the person. Whether that clears the originality bar is genuinely uncertain, and a reasonable argument exists that it does not. By contrast, where a human makes substantive creative decisions — selecting specific phrasing, restructuring content, adding original analysis, or exercising real editorial judgment over what stays and what goes — there is a considerably more defensible originality claim, because the human's own skill and judgment are now embedded in the final work.

It helps to think of AI-involved content as sitting on a spectrum rather than falling into a single category.

At one end: a marketing image generated from a single short prompt — say, "a lawyer shaking hands with a client, professional style" — where the output is used exactly as generated, with no further human editing, selection among alternatives, or creative direction. Here, the human's contribution is largely an idea or instruction rather than an expression. This is the weakest position for a copyright claim, and under the EBC v. Modak standard, arguably falls short of the "modicum of creativity" the law expects from the human side.

At the other end: a blog post where an AI tool is used to generate an initial outline or rough first draft, but a human writer substantially rewrites the language, restructures the arguments, adds original examples and analysis, and makes deliberate editorial choices about tone and content. Here, the final expression carries meaningful human authorship, even though AI played a role in the process. This is a considerably stronger position — arguably no different, in principle, from a writer who works from a research assistant's notes, or a photographer who relies on editing software.

Most real-world use sits somewhere between these two poles. The practical lesson is that the amount and nature of human creative involvement — not merely the fact that AI was used somewhere in the process — is what will likely determine whether a resulting work can be protected.

Even where a work clears the originality bar, a second question follows: who owns the copyright in it? Under Section 2(d)(vi), the answer is framed as "the person who causes the work to be created" — but this is not necessarily the same person who typed the prompt.

Consider a business where an employee uses an AI tool to generate content as part of their job. Ordinary copyright principles around works created in the course of employment would likely still apply, meaning the employer may hold the resulting rights, much as it would for any other work an employee produces during employment. Where a freelancer or agency is contracted to produce AI-assisted content, the contract terms — not the AI tool — typically govern who owns the final output. This is exactly why clear IP assignment clauses in vendor and freelancer agreements matter as much for AI-assisted work as for any other commissioned work.

There is also a separate layer that is easy to overlook: the terms of service of the AI tool itself. Different AI platforms take different contractual positions on who owns generated output, whether the platform retains any rights to reuse it, and what restrictions apply to commercial use. These terms operate independently of copyright law — a platform's terms of service cannot override the Copyright Act, but they can still affect what you are contractually permitted to do with the output, or what rights you may have implicitly granted back to the platform. Reading the AI tool's terms before relying on its output commercially is a basic, often-skipped step.

Businesses sometimes ask whether they can simply register AI-assisted content with the Indian Copyright Office and treat the certificate as settling the question. It is worth being cautious here.

Copyright registration in India, as with most works, is not compulsory for protection to exist, and registration itself is not an automatic guarantee that a court will later find the work original or properly authored — registration creates a public record and useful evidentiary support, but the underlying legal questions about originality and authorship of AI-assisted content remain open regardless of whether registration is granted. Public guidance from the Copyright Office specifically addressing AI-generated works is limited, and practice in this area is still developing, so it would be inaccurate to claim a settled, official position exists on exactly how AI involvement should be disclosed or evaluated at the registration stage.

Our practical recommendation is not to leave this to guesswork. When registering copyright for a work that involved AI tools at any stage, it is sensible to be able to clearly describe — and where possible document — the human creative contribution: drafts, revision history, prompt logs, and selection decisions, rather than presenting a purely AI-generated output as if it were wholly hand-created. Being transparent and prepared is a stronger long-term position than hoping the question never comes up, particularly if the work later becomes commercially important enough to be challenged or litigated.

India is not alone in facing this issue, and it is worth noting the broader context without overstating expertise in foreign law. In the United States, the U.S. Copyright Office has taken a comparatively restrictive public position, generally declining to register works where the creative expression was produced by an AI system without sufficient human authorship, while indicating that works combining meaningful human-authored elements with AI-generated material may be registrable for the human-authored portions. Other jurisdictions are approaching the question differently, and there is no international consensus yet.

This global picture matters for two reasons. First, it confirms that the uncertainty in India reflects a genuinely unsettled area of law worldwide, not a gap unique to Indian legislation — so there is no simple "settled" foreign rule to import wholesale. Second, if your business operates internationally or licenses content across borders, the copyright status of the same AI-assisted work may be assessed differently in different countries. A foreign copyright office's position has no binding effect in India, and vice versa.

For Indian creators and businesses, the safest posture is to treat foreign guidance as useful context for where the law may be heading, not as legal authority governing your rights here. Each jurisdiction where you publish, license, or enforce content should be evaluated on its own terms.

There is a practical risk that often gets overlooked in the rush to use AI tools to move faster: if your AI-generated content is not copyrightable, that cuts against you as much as it does against anyone else.

Copyright is what allows a business to stop a competitor from copying its website content, marketing copy, or graphics wholesale. If a piece of content lacks the human originality needed for protection, it is not just that you cannot register it — it may mean no one has an enforceable exclusive right over it at all, including you. A competitor could, in principle, copy the same AI-generated marketing image or blog paragraph, and you would have no independent copyright claim to stop them, precisely because the work was never protected in the first place.

This is a meaningful strategic consideration, not just a legal technicality. If a piece of content is central to your brand identity, your sales funnel, or your competitive positioning, it deserves more human creative investment than a single unedited AI prompt — not only to make it better, but to make it legally yours in a way a court is more likely to recognize. Reserve purely AI-generated, unedited output for lower-stakes, disposable content, where the absence of strong legal protection is an acceptable trade-off for speed.

Until Indian courts or the legislature provide clearer guidance, treat AI-assisted content as an area requiring active management rather than passive assumption of protection. A few practical, low-cost steps meaningfully improve your legal position:

**Document human creative contribution.** Keep drafts, revision history, prompt records, and notes on editorial decisions for any AI-assisted work you consider commercially important — this is exactly the evidence that matters if originality is ever challenged.

**Review the AI tool's terms of service before commercial use.** Different platforms take different positions on output ownership, commercial-use restrictions, and rights the platform itself retains. Do not assume you own, or can freely commercialise, output simply because you generated it.

**Do not assume default copyright protection for raw, unedited AI output.** If content is genuinely important to your brand or business, invest real human editorial effort into it rather than publishing AI drafts as final work.

**Triage your content by importance.** Reserve careful human involvement for high-value assets — core website copy, brand assets, flagship content — and accept weaker protection for lower-stakes, high-volume content where speed matters more than exclusivity.

**Get specific legal advice before high-value AI-assisted launches.** If an AI-assisted work is central to a product launch, funding round, or licensing deal, a brief legal review of authorship and documentation is far cheaper than a later dispute.

The uncertainty around AI-generated content does not mean you should leave your copyright position to chance. If your business publishes AI-assisted blog content, marketing material, designs, or code, it is worth getting specific advice on two fronts: whether a particular work is likely to qualify for copyright protection, and how to document the human creative contribution in a way that strengthens your position if it is ever challenged.

Online Legal Mitra assists creators and businesses with copyright registration across literary, artistic, and software works, and can advise on how to properly document human authorship for AI-assisted content before you register or rely on it commercially. We also help review AI tool terms of service where relevant, and advise on IP assignment clauses for teams and freelancers producing AI-assisted work.

If you are unsure whether your AI-assisted content is worth registering, or how to strengthen a weak claim before it becomes a problem, visit our copyright registration services at /services/copyright/copyright-registration or contact us directly at +91-8959420521 or info@onlinelegalmitra.com for a consultation. Since 2014, our team in Gwalior has helped businesses and individual creators across India navigate exactly this kind of evolving, unsettled legal terrain — with honest advice about what is actually protectable, not false certainty.