AI and Trade Mark Squatting Guidelines in India

Published 22 October 20268 minute read

India is about to get AI guidelines for the Trade Marks Registry. The Controller General of Patents, Designs and Trade Marks has invited comments on rules for AI use and trade mark squatting. The window is three weeks from the public notice of 16 October 2026.

The notice follows an order of the Delhi High Court in Stryten Energy LLC v Nidhi Agarwal, W.P.(C)-IPD 16/2026. A draft titled "Guidelines on Use of Artificial Intelligence in Proceedings before the Trade Marks Registry, 2026" is published with it. Anyone who files, opposes or defends a trade mark in India should read it.

What the Delhi High Court ordered

The petition is still pending. On 16 April 2026 Justice Tushar Rao Gedela directed the Registry to furnish draft guidelines to check AI generated pleadings. The concern was the citation of false and fake case law before the Registrar.

On 13 October 2026 Justice Purushaindra Kumar Kaurav went further. The court said there have to be comprehensive guidelines on the use and misuse of AI and on the "menace of squatting". It directed the CGPDTM to invite suggestions and objections within three weeks of receiving the order. The office must then compile them and suggest workable guidelines within a further three weeks. The matter is listed on 14 December 2026.

Key points

Comments go by email to llc-ipo@gov.in with the subject "Suggestions/Objections – Guidelines on AI and Squatting".
The window is three weeks from publication of the notice dated 16 October 2026. That points to about 6 November 2026.
The office says the final guidelines will apply to all pending and new trade mark applications.
The draft requires a signed declaration on AI use with every submission.
AI must not be used to write affidavits or to create or alter evidence such as invoices or proof of use.
The published draft deals with AI only. It contains no rules on squatting yet.

What the draft AI guidelines require

The draft applies to every proceeding under the Trade Marks Act, 1999. That covers filing, examination, opposition, rectification, cancellation, renewal, assignment and recordal. It would come into force on notification and apply to pending matters as well as new ones.

A declaration with every submission

Clause 8.3 requires one declaration in the Annexure I form with every submission. It must say whether AI was used. If it was not, a simple statement is enough. If it was, the declarant must confirm that every citation, statutory reference, legal proposition and material fact suggested by AI was checked independently. The party and the authorised representative must both sign where an agent or advocate files.

Clause 5.2 makes disclosure absolute. Failing to disclose AI use is treated as a misrepresentation to the Registry.

Verification from recognised sources

Each citation must be checked against "Authorised and Recognised Non-AI Legal Sources". The list covers court websites, the Official Gazette, certified copies, the Registry's own portal and established law reports and databases. A citation taken from an AI tool does not count as verified until it is cross checked against one of these.

The representative must confirm that each case exists, that the citation details are right, that the holding is stated accurately and that the case is relevant. A dispute about relevance alone will not trigger action. On the holding and relevance checks, action lies only where a case does not exist or is deliberately or recklessly cited for a proposition it does not contain.

No AI in affidavits or evidence

Clause 14 is the most important rule for brand owners. AI must not be used to write or alter the substance of an affidavit, declaration or statement of personal knowledge. It must not be used to create or manipulate evidence. The list includes invoices, sales data, labels, advertisements, website extracts, market surveys, translations and proof of use.

Where authenticity is in doubt, the Registrar may act under section 127 of the Act after a hearing. The Registrar may direct production of the native file, metadata, audit trails and creation history. Material that is not shown to be authentic may be struck off, and an adverse inference may be drawn.

Consequences and reporting

The Registrar may start proceedings suo motu or on a party's request. A show cause notice and a hearing come first. The Registrar may then strike out the false citation or the whole submission. Agents may be referred for disciplinary action, and advocates to the Bar Council.

Any party may report a suspected fake citation or false declaration in the Annexure II form. Reports made in good faith carry no adverse consequence. Frivolous reports can themselves lead to action. A pending report does not stay the proceeding or extend any deadline.

The draft also binds the office. Examiners and Registry officers may use AI only as an aid. They must not rely on any AI generated proposition or precedent unless they have verified it.

What the draft does not cover

The published draft says nothing about squatting. The court asked for guidelines on both subjects, and the public notice invites comments on both. No squatting text exists yet. The comments received may shape what the office proposes to the court.

The Act already gives some tools. Under section 11(10)(ii) the Registrar must take into account bad faith by the applicant or the opponent. Oppositions under section 21 and rectification under sections 47 and 57 remain available. What is missing is a clear office practice for spotting bulk or speculative filings at examination. This is where comments from brand owners can help.

The draft also leaves some practical questions open. It does not say how the declaration will be filed online, or what happens to a submission filed without one in a pending matter. Its definition of AI is broad and includes natural language processing tools. It is unclear whether routine translation or spell checking triggers the fuller declaration. These points are worth raising in comments.

How India compares with the US, Europe, the UK and China

United States. The USPTO's guidance of 11 April 2024 (89 FR 25609) covers AI tools in patent and trade mark practice. It imposes no general duty to disclose AI use. It relies on the signature and certification rule in 37 CFR 11.18 instead. A signer must check every citation, and simply relying on an AI tool is not a reasonable inquiry. The guidance warns against filing AI generated specimens that do not show real use. The US has no bad faith refusal ground at examination. Speculative filings are attacked through lack of a bona fide intent to use, and through expungement and reexamination added by the Trademark Modernization Act of 2020.
European Union. The EUIPO has no published rule requiring disclosure of AI use in filings. On squatting, Article 59(1)(b) of the EU trade mark regulation allows invalidation of a mark filed in bad faith. In Sky v SkyKick (C-371/18, 2020) the Court of Justice held that filing without intent to use can be bad faith. This applies where the applicant meant to undermine third parties or obtain rights for purposes other than those of a trade mark.
United Kingdom. The UK has no AI declaration rule for trade mark proceedings either. In Pro Health Solutions (BL O/0559/25, June 2025) the Appointed Person dealt with false citations from a party who used ChatGPT. It also dealt with real cases cited by a trade mark attorney for propositions they did not clearly support. He suggested that the Registrar warn parties about AI risks when asking for submissions. Bad faith under section 3(6) of the Trade Marks Act 1994 is the main UK tool against squatting. The UK Supreme Court applied it in SkyKick in 2024.
China. Article 4 of the Trademark Law, amended in 2019, requires rejection of bad faith applications not intended for use. This is the most direct anti squatting rule among the major offices.

If adopted, India's draft would be stricter than the US and UK rules on AI. It would add a mandatory declaration to the existing duty to file accurate material. It also deals with AI generated evidence in more detail than the US guidance does. Our earlier note on AI guidelines for patent examination in India covers the parallel rules on the patent side.

Practical steps for applicants and brand owners

Send comments within three weeks of publication of the notice, which is likely by 6 November 2026. Squatting needs input most, as no draft text exists yet.
Set an internal rule that every citation in a reply, counter statement or written submission is checked against the full judgment.
Keep native files and metadata for invoices, labels, screenshots and other evidence of use.
Do not use AI to draft affidavits of use. Use it, if at all, only for formatting or language checks, and record that use.
Ask your agent how they will handle the declaration once the guidelines are notified.
If a rival's filing looks speculative or copied, consider opposition and plead bad faith under section 11(10)(ii).

Our trade mark services cover filing, oppositions and rectification in India and abroad. For the identity checks that now apply to filers, see our note on trade mark eKYC in India.

Frequently asked questions

Do I have to disclose AI use to the Trade Marks Registry in India?

Not yet. The draft guidelines would require a signed declaration with every submission once they are notified. Until then, the duty is to ensure that everything filed is accurate.

Can I use AI to prepare a trade mark reply in India?

Under the draft, yes, if a human verifies every citation and fact and the declaration is filed. AI must not be used to write affidavits or to create evidence.

How do I comment on the AI and squatting guidelines?

Email llc-ipo@gov.in with the subject "Suggestions/Objections – Guidelines on AI and Squatting" within three weeks of the 16 October 2026 notice.

AI guidelines for the Trade Marks Registry and trade mark squatting