IT Ministry Plans AI-Based Fraud Detection System for Investor Refund Applications

The government plans to introduce 12 AI services to detect fraud in nearly 2.9 lakh pending investor refund claims across India.

The National e-Governance Division (NeGD), under the Ministry of Electronics and Information Technology, is seeking to deploy artificial intelligence tools to help the Investor Education and Protection Fund Authority (IEPFA) identify fraudulent applications involving unclaimed shares, dividends and matured deposits.

The IEPFA, which operates under the Ministry of Corporate Affairs, is currently examining approximately 2.9 lakh refund forms through a largely manual process. The proposed system would verify applicants’ identities, examine their entitlement to the claimed assets and identify suspicious patterns before refunds are authorised.

FCRF’s Chief AI Officer Certification Returns on November 14 to Prepare Future AI Leaders

What Types of Fraud Has the Government Identified?

According to a request for quotation (RFQ) issued by NeGD, the proposed system will examine approximately 28 reported and anticipated fraud patterns associated with investor refund applications. These include forged identity documents, altered entitlement letters and claims submitted using manipulated financial information. Authorities have identified instances in which entitlement letters were uploaded again after the amounts mentioned in them had been changed.

The proposed checks will also cover claims involving deceased investors. These include falsified death certificates, forged signatures on no-objection certificates and attempts to conceal other legal heirs who may have a legitimate entitlement to the assets. The authority has also reported multiple claims against the same securities, raising concerns about fraudulent or overlapping applications.

How Will AI Identify Suspicious Refund Applications?

The proposed services will combine document verification with analysis of information across different claims. AI tools will compare information submitted by applicants against available official records, match names across documents and identify applications connected through common identifiers.

Suspicious patterns may include a single mobile number linked to multiple PANs, identical email addresses appearing in unrelated claims and the same bank identification code being used by apparently unrelated applicants.

The RFQ also identifies a newer concern involving repeated claim explanations potentially generated using large language models. Such applications may contain similarly worded statements submitted in large numbers. These checks are intended to help officials distinguish legitimate refund requests from applications requiring further examination.

Will AI Have the Power to Approve or Reject Claims?

The government has made clear that artificial intelligence will support the examination process rather than replace official decision-making. The proposed system will prepare summaries of individual claims and recommend relevant noting codes used by the authority. It may also generate scores to help officials assess applications.

However, the RFQ explicitly states that “the score is informative”, with responsibility for action remaining with the designated government officer. AI services will not independently approve or reject refund applications, determine their status or formally establish that fraud has occurred.

This distinction keeps final responsibility for decisions with human officials, even where automated checks identify potential irregularities.

When Will the Proposed AI System Be Introduced?

NeGD published the RFQ on October 13, with proposals required to be submitted by November 5. The proposed implementation period is four months, followed by an initial 12-month operations and maintenance arrangement. The contract also provides options for two additional extensions of 12 months each.

The initiative is intended to strengthen scrutiny of investor refund claims by bringing together identity verification, document examination and cross-application analysis.

For investors seeking the return of unclaimed financial assets, the proposed changes would introduce additional checks into a process currently dependent largely on manual examination.

The central objective is to identify potentially fraudulent claims while retaining human oversight over decisions involving legitimate investor entitlements.