AI in Patent Examination: A New Framework for Technology-Assisted Examination

House no. 316, Ground Floor "A" Sector, Yelahanka New Town Bengaluru- 560 064

1203, The Spire-2, Shital Park BRTS, 150 Feet Ring Road,Rajkot - 360007 Phone : +91 281 242 731

203, 2nd Floor, Shriji Palace,Savsar Plot, Main Road, Morbi-363641 INDIA Phone:91 2822225263

312-313, 3rd Floor,
Abhishek Complex, Akshar Chowk, Old Padra Road,
Vadodara 390 020 INDIA
Phone: (0265) 2322015

On 7 August 2026, the Indian Patent Office published its Guidelines for the Use of Artificial Intelligence in Patent Examination Procedures.

This is an important development as it formally recognizes and regulates the use of AI and Generative AI by Patent Office examiners and controllers.

The Guidelines adopt a “human-in-the-loop” approach. AI may assist examiners in a range of activities, but substantive statutory decisions remain the responsibility of the examiner or controller.

Permitted or potentially permitted assistive uses include:

• IPC/CPC classification;
• Identifying search terms and concept clusters;
• Preliminary claim-feature extraction;
• Translation assistance;
• Structuring or improving draft communications;
• Identifying technical and legal references;
• Preliminary novelty and inventive-step analysis;
• Identifying potential clarity issues;
• Prior-art searching through approved tools;
• Preliminary assessment of sufficiency of disclosure; and
• Preliminary research concerning case law and legal concepts.

However, the Guidelines contain important safeguards. AI-generated material cannot replace the examiner’s independent assessment, and AI-generated citations and references must be independently verified.

Of particular significance, unpublished patent application material, confidential Office records and internal deliberative material may not be entered into public AI tools. The Guidelines also restrict the issuance of FERs, hearing notices, decisions and other official communications solely on the basis of unverified AI-generated material.

The Guidelines also contemplate governance mechanisms, including recording of material AI use, an AI Governance Committee, training, quality review and incident reporting.

Potential impact on patent prosecution

For applicants and patent practitioners, this development is significant for two reasons.

First, AI-assisted searching may enable examiners to process larger volumes of prior-art material and identify technically related documents more efficiently.

Second, the Guidelines themselves recognize risks associated with model sensitivity, prompt sensitivity and inaccurate or fabricated outputs. Accordingly, an AI-generated search result or technical comparison should not automatically be assumed to reflect a definitive assessment of patentability.

This may become particularly relevant when responding to FERs where a cited reference appears only tangentially related to the claimed invention.

Our view: Applicants should continue to emphasize precise claim drafting, comprehensive technical disclosure and strong distinction from prior art. At the same time, practitioners should carefully scrutinize AI-assisted search results and verify every substantive citation relied upon by the Patent Office.

Disclaimer: This update is intended for general informational purposes and does not constitute legal advice. The Patent and Trade Marks Manuals and the pharmaceutical and biotechnology examination guidelines discussed above are drafts and may be amended before finalization. The applicable legislation, Rules, judicial decisions and final Office practice should be considered when advising on any specific matter.

Sources: Indian Intellectual Property Office (IP India); International Trademark Association; publicly available commentary on the 2026 draft materials.