Medianama reports AI-assisted hackers diverted Rs 65 cr from Delhi-NCR NBFCs
Medianama, citing Indian Express and Delhi Police, reports hackers diverted Rs 65 crore from 12 NBFCs; check disbursal APIs against the CERT-In advisory.
- Published
- Exploitation
- Yes, per Delhi Police via Medianama (12 NCR cases, 20 arrests) and CERT-In, which says it observed campaigns; amounts not independently verified
What happened
According to a Medianama report published on November 18, 2026, which relays an Indian Express story, hackers possibly using AI diverted approved loans at 12 non-banking financial companies (NBFCs) in Delhi-NCR over three months and took over Rs 65 crore. Medianama says the Indian Express cited senior Delhi Police officers, who described these cases as part of more than 150 first information reports filed by NBFCs across India. These figures are the police account as relayed by Medianama. The sources do not say they were independently verified, and they do not say whether the affected companies have confirmed the cases.
Delhi Police told the Indian Express, as relayed by Medianama, that the attackers found loans that were approved but not yet disbursed and redirected the money to mule accounts. Gaurav Tyagi, DCP at Intelligence Fusion and Strategic Operations, is quoted as saying that traces of AI use were detected and that the incidents are largely attributable to weaknesses in the public-facing cyber systems of NBFCs. Police also said malware is typically used against the front end of NBFC systems and could stay in place for over 7-8 days.
Medianama says police described one mechanism: an NBFC’s firewall whitelists the outbound IP addresses of a cloud data centre, and attackers can send disbursal commands from addresses the firewall already trusts. The command carries a mule account in place of the borrower’s account. This is the police account as reported, not a finding verified by Medianama.
Medianama reports that police arrested 20 people in the 12 NCR cases, all linked to creating or procuring mule accounts, and that Rs 18 crore of the Rs 65 crore was seized before it left those accounts. It says the Indian Express named only two companies. Hero FinCorp reportedly lost Rs 12 crore on August 17. Unifinz Pvt Ltd claimed, in a complaint dated October 21, to have lost Rs 12.49 crore in 550 transactions over two days. The other entities are not named, and the sources give no confirmation from any company beyond what is attributed to the Unifinz complaint.
The official CERT-In advisory CIAD-2026-0046, originally issued on October 9, 2026 and rated Critical, says CERT-In/CSIRT-Fin has observed multiple campaigns against NBFCs, digital lending platforms, fintechs, payment service providers and wallet operators. It says attackers steal payment gateway and partner-bank API credentials and call disbursal or fund-transfer APIs directly from the entities’ own whitelisted infrastructure, bypassing application-level checks such as KYC, sanction and per-transaction limits. CERT-In says its analysis indicates the possible use of AI-assisted offensive tooling. The advisory does not mention the Delhi-NCR cases, so the link between them is not confirmed.
Am I affected?
CERT-In addresses the advisory to NBFCs, digital lending platforms, fintechs, payment service providers and wallet operators. You may be exposed if you have any of the following, which CERT-In lists as initial access routes:
- Internet-facing applications or application server middleware with known vulnerabilities.
- Deprecated, unauthenticated or unused APIs, including ones with unrestricted file upload.
- Long-lived static API tokens hard-coded in source code.
- Plaintext credentials in configuration files, source code or databases.
- Applications or remote services without MFA.
- Source-code repositories at risk of forged commits.
CERT-In also lists signs seen after compromise: web shells and backdoors, deleted or truncated logs, tampered EDR tools, and payouts that carry dummy reference numbers with no matching record.
Not confirmed in the sources: confirmation by the affected companies, the exact entry point in the 12 NCR cases, which malware was used, which AI tools were used, who the attackers are, and the identities of the unnamed companies. Medianama also notes that the CERT-In advisory does not name affected entities or give incident counts or amounts stolen.
What to do now
The steps below come from the CERT-In advisory.
- Enforce MFA on payment platforms, administrative portals, payment and bank APIs, and remote access services.
- Keep a complete API inventory, decommission deprecated and unused APIs, enforce per-session authentication, and eliminate long-lived static tokens.
- Store API keys and credentials in a secret management solution or HSM and rotate them.
- Require maker-checker or second-level verification for all fund movement. Validate every payout against approved records and reject reference identifiers with no matching record.
- Enforce per-transaction and per-day ceilings and beneficiary validation (penny-drop, name-to-account matching) at the API or payment stage.
- Restrict outbound traffic from production servers through allow-listing, and monitor transfers in near real time for velocity, amount and new-beneficiary anomalies. Reconcile your records with bank and aggregator statements.
- Keep centralized logs for at least 180 days and deploy EDR with tamper protection on servers and endpoints.
- If you see suspicious activity: disable compromised accounts, suspend affected APIs, preserve forensic evidence, rotate all credentials, and report to CERT-In/CSIRT-Fin at incident@cert-in.org.in with relevant logs.