AI Sparks Patent Revolution in India: New Rules, Inventorship Limits, and Compliance Challenges
CRI Guidelines Clarify AI-Assisted vs. AI-Generated Inventions Under Section 3(k)
IPO Refuses AI Inventorship — Human Contribution Remains Central to Patent Law
By Vishwas Kumar
New Delhi: May 27, 2026:
India’s patent regime is undergoing a major transformation as artificial intelligence reshapes innovation. Recent amendments to the Patents Act, 1970, revised Computer Related Invention (CRI) guidelines (2025), and the refusal to recognize AI systems like DABUS as inventors (April 2026) highlight how India balances technological progress with legal safeguards. Background
India’s technology hubs are witnessing a surge in AI-driven innovation. Executives argue that AI is turbocharging patent creation, but the legal framework has tightened. The Indian Patent Office (IPO) has issued revised CRI guidelines (2025), amended Patent Rules (2026), and clarified inventorship requirements under the Patents Act, 1970. These changes directly affect startups, R&D centers, and multinational corporations filing AI-related patents.
The Supreme Court judgment in Manoj Kumar vs Union of India is an important decision dealing with constitutional rights, administrative fairness, and the legality of government actions affecting citizens. The case examines the scope of judicial review, procedural compliance by public authorities, and the balance between executive powers and individual protections guaranteed under the Constitution of India. The Supreme Court emphasized that governmental decisions must satisfy standards of transparency, reasonableness, and non-arbitrariness, making this judgment highly relevant for constitutional law, administrative law, and public policy research.
Key Legal Developments
- Section 3(k) of the Patents Act, 1970
- Bars patents for “computer programs per se, algorithms, or mathematical methods.”
- Revised CRI guidelines (2025) introduced the “technical effect” test: AI-assisted inventions are patentable if they produce a demonstrable technical effect (e.g., improved hardware efficiency, medical diagnostics).
- Purely autonomous AI outputs remain non-patentable. globallawexperts.com
- Inventorship and the DABUS Case (2026)
- On 15 April 2026, the IPO rejected an application naming DABUS (Device for the Autonomous Bootstrapping of Unified Sentience) as inventor.
- Reasoning: Sections 6 and 7 of the Patents Act require inventors to be “persons,” interpreted as natural or legal persons.
- India joins the US, UK, EU, and Australia in refusing AI inventorship.
- Patent Rules Amendments (2026)
- Form 27 working statements: patentees must annually disclose whether their invention is being worked in India.
- Non-compliance can trigger compulsory licensing inquiries and penalties.
- This is especially relevant for AI delivered as cloud services, where “working” in India is harder to demonstrate. globallawexperts.com
- International Filing Strategy
- PCT reforms (2026) altered search timelines and fee structures.
- For companies targeting limited jurisdictions, an India-first filing followed by Paris Convention filings may now be more cost-effective than the PCT route.
Implications for Industry
- Startups must carefully draft claims to highlight human inventive steps.
- Multinationals need to align global filings with India’s stricter CRI guidelines.
- Compliance risks: Failure to file Form 27 correctly can jeopardize patent rights.
- Policy balance: India aims to encourage innovation while preventing monopolization of abstract algorithms.
Comparison Table: AI Patent Rules in India
|
Aspect |
Old Framework |
Current Framework (2025–26) |
|
Section 3(k) |
Broadly excluded computer programs |
Technical effect test allows AI-assisted inventions |
|
Inventorship |
Ambiguous |
Only natural/legal persons; AI excluded |
|
Form 27 |
Limited disclosure |
Expanded obligations; penalties for non-compliance |
|
International Filing |
PCT preferred |
India-first + Paris Convention viable |
Detailed FAQ
Q1: Can AI be listed as an inventor in India?
No. Only natural persons qualify under Sections 6 and 7 of the Patents Act. AI systems like DABUS cannot be inventors.
Q2: Are AI-assisted inventions patentable?
Yes, if they demonstrate a technical effect beyond algorithms per se. For example, AI improving medical imaging hardware may qualify.
Q3: What is Section 3(k) and why is it important?
Section 3(k) excludes computer programs, algorithms, and mathematical methods from patentability. It is the main hurdle for AI-related claims.
Q4: What is Form 27 and why does it matter?
Form 27 is a mandatory annual disclosure of whether a patent is being worked in India. Non-compliance can lead to compulsory licensing.
Q5: How should companies file AI patents internationally?
With PCT reforms, India-first filing followed by Paris Convention filings may be more cost-effective for limited jurisdictions.
In summary: India’s patent law now clearly distinguishes between AI-assisted inventions (patentable) and AI-generated inventions (non-patentable). The refusal to recognize AI inventorship and stricter compliance rules mark a decisive shift, ensuring that human creativity remains central to intellectual property protection while adapting to the AI era.