India's national cyber security agency, the Indian Computer Emergency Response Team (CERT-In), issued a critical-severity advisory on 20 November 2026 warning that fraudsters are making unauthorised payments by quietly changing the business records that decide who gets paid. The advisory, numbered CIAD-2026-0062, also says attackers are using AI-driven tools that can run parts of an intrusion on their own. MapMyTimes reported the advisory when it was issued; this explainer sets out what it means in plain terms.

What is payment workflow fraud?

Payment workflow fraud is when an attacker does not break into the payment system itself but changes the records behind it, so that the normal, approved process sends money to the wrong place. CERT-In says the attackers it observed worked through business applications and workflows run by service providers, altering records or approval relationships that determine the beneficiary. The payment then goes out through trusted systems and standard approvals, which is why, in the agency's words, it may not raise immediate suspicion.

Think of a company that pays hundreds of suppliers each month from a list of vendor bank accounts. If someone with access to that list swaps one supplier's account number for their own, every check on the payment request itself can pass, because the request matches the record. The record is what was tampered with.

Why is it in the news?

It is in the news because CERT-In rated the threat critical and said it is happening outside the banking sector. The advisory names technology service providers, IT-enabled services firms and non-financial payout platforms, including those that support government-linked payment processes, as affected entities. It says the weak points are gaps in independent verification, reconciliation and review of changes to business records.

The advisory did not say how much money was lost, which organisations were hit or who was behind the activity, and it has not said whether investigations are complete.

What are AI-driven agentic offensive frameworks?

They are attack toolkits that use AI agents to carry out several steps of a cyber intrusion with less human effort. CERT-In calls them agentic offensive frameworks and says its threat hunting found them being used against enterprises, technology and service providers, government bodies, and academic and research institutions.

According to the advisory, the techniques seen include:

  • automated scanning to find exposed systems and possible routes in;
  • quickly building or adapting exploits for known software flaws, including by studying software updates;
  • several agents working in parallel on reconnaissance and privilege discovery;
  • changing scripts and attack steps in response to errors or blocked attempts;
  • using valid passwords, legitimate admin functions and trusted connections to go deeper.

CERT-In says this cuts the manual work needed and can shorten the time between first scanning and later stages of an attack. The concern is not limited to India: earlier this week MapMyTimes reported that the European Commission confirmed an AI-assisted cyberattack on its staff systems.

What does CERT-In want organisations to do?

CERT-In wants organisations to check the record and the authority behind a payment, not just the payment request. For government bodies and other owners of services run by outside providers, it asks them to:

  • identify which provider accounts and integrations can change payment records or approve transactions, and restrict them;
  • reconcile payment instructions against an independent source that cannot be changed through the same provider system;
  • require separate verification for high-risk changes such as new beneficiaries or bank details, since several approvals inside one compromised system may not be enough;
  • agree with providers in advance on log access, investigation and suspending payments if something goes wrong.

Service providers and payout platforms are told to keep client environments separate so that access for one client cannot reach another, and to watch changes to payment records alongside admin activity and the payments that follow, even when every step was done by an authorised account. Organisations generally are asked to review internet-facing applications, privileged accounts and logs for unusual scanning, credential use and outbound data transfers.

How does it affect you?

It affects you mainly if you run, manage or work in finance or IT for a business, government office or institution that pays vendors, staff or beneficiaries through a software platform or outside provider. For such teams, the practical lesson is that a change to a beneficiary's bank details deserves a call-back or other check through a separate channel before money moves.

For individuals, the advisory is a reminder of where to go if money leaves an account without permission. CERT-In says unauthorised financial transactions should be reported promptly on the National Cyber Crime Reporting Portal or by calling the 1930 helpline, and that law enforcement, regulators and payment authorities should be told as applicable.

Organisations covered by CERT-In's directions of 28 April 2022 must report applicable cyber incidents to the agency within six hours of noticing them or being told about them, with logs sent to incident@cert-in.org.in.

What happens next?

What happens next depends largely on the organisations named in the advisory. CERT-In has pointed them to its earlier advisories CIAD-2026-0046, CIAD-2026-0049 and CIAD-2026-0058, and to a blueprint on defending against AI-assisted exploitation of vulnerabilities (CISG-2026-02). The advisory page shows it was last updated on 22 November 2026, so further revisions are possible. Details such as losses, affected entities and any arrests have not been made public.

Photo source