CERT-In warns of AI-driven attacks that quietly change business records to steal money through real payments

India's cyber security agency says attackers are now altering approval records and vendor data so that fraudulent payments pass normal checks, including at firms that handle government-linked payments. Here is how the attack works and how to protect your business.

A server room. File photo.
A server room. File photo. · Photo: BalticServers.com / Wikimedia Commons, CC BY-SA 3.0

Key points

  • CERT-In has warned of AI-driven cyberattacks on business payment systems across sectors
  • Attackers change business records or approvals, so fraudulent payments pass normal checks
  • Technology firms supporting government-linked payments and vendor payouts are affected
  • CERT-In has seen an 'AI-driven agentic offensive framework' used against enterprises and government bodies
  • It advises all sectors to review exposure, access controls and monitoring

India's national cyber security agency, CERT-In, has warned that hackers are using AI-driven tools to steal money through business payment systems. Instead of breaking into payment gateways, they quietly change the records and approvals that decide who gets paid.

Key points from the advisory

  • Target: business payment systems in every sector, not just banks.
  • Also hit: technology firms that support government-linked payments and vendor payout platforms.
  • Method: altering vendor details, invoices or approval records so fake payments look genuine.
  • Tool: an "AI-driven agentic offensive framework" seen attacking companies, government bodies and universities.
  • Advice: review exposure, access controls and monitoring.

How the attack works

A company's payment system usually trusts its own records. If attackers change a supplier's bank account number, approve a fake invoice, or tweak an approval step, the payment goes through the normal, legitimate process, and the money lands in the attacker's account. According to the advisory, such transactions can pass the usual application and approval checks.

Rows of servers in a data centre
Attackers increasingly target the business records behind payments, not just payment systems. File photo. Photo: Christopher Bowns / Wikimedia Commons, CC BY-SA 2.0

Why AI makes it worse

AI "agents" can scan systems, find weak points and act faster than human hackers. Earlier in November, CERT-In had warned that attackers were using AI to break into the live systems of NBFCs, fintech companies and payment aggregators.

How businesses can protect themselves

  • Verify any change in a supplier's bank details by phone, using a number you already have.
  • Require two people to approve large or unusual payments.
  • Turn on multi-factor login for finance and ERP systems.
  • Keep logs of who changed records, and review them regularly.
  • Report incidents to CERT-In at incident@cert-in.org.in.

More on The World Brief

Sources

Read next