Cert-In, I4C warn financial firms of AI-driven cyberattacks, unauthorised fund transfers

The Indian Computer Emergency Response Team (Cert-In), which deals with cyber security incidents, has warned banks, non-banking financial companies (NBFCs) and fintech firms about a campaign of AI-driven cyberattacks that has led to unauthorised fund transfers worth crores. In two advisories issued on November 3 and 4, Cert-In and the Indian Cyber Crime Coordination Centre (I4C) flagged attacks in which hackers are using AI agents to break into financial institutions’ systems and exploit payment and fund-disbursement platforms to transfer money without authorisation.Hackers were using AI agents to break into systems and exploit payment and fund-disbursement platforms. (Representative photo)The November 3 advisory, issued by I4C’s National Cybercrime Threat Analysis Unit (NCTAU) along with Cert-In, said attackers were exploiting unpatched software and incorrectly configured internet-facing servers to gain access to financial institutions’ systems. They then target automated fund-disbursement systems and application programming interfaces (APIs), which are used to process payments and transfers.The money is transferred to accounts, known as mule accounts, across different banks and quickly moved or withdrawn through IMPS and NEFT. Attackers also delete records to make investigations difficult, the advisory said.The November 4 advisory, issued by Cert-In’s Computer Security Incident Response Team-Finance (CSIRT-Fin), described a further escalation of the campaign and identified the threat actor as “CRAFTY FALCON”. The agency said its analysis of incidents reported by NBFCs, digital lending platforms, fintech companies, payment aggregators, payment service providers and cryptocurrency exchanges showed that attackers were using publicly available AI-based offensive security tools to carry out attacks at high speed.These include ARTEX, an AI-based tool, the PentAGI autonomous multi-agent penetration testing system, the Cairn attack search engine and the CyberStrikeAI framework.The November 4 advisory said these tools can generate large numbers of customised attack scripts, change their methods based on a target’s response and carry out several stages of an attack with limited human involvement.Cert-In said some financial institutions that received earlier restricted advisories and took corrective steps were targeted again. Attacks were sometimes timed to coincide with weekends, holidays and other periods when security teams had fewer staff monitoring systems.The agency flagged methods used by attackers, including breaking into cloud systems, moving from testing environments into production systems, stealing credentials and exploiting weaknesses in payment APIs. It warned that attackers were targeting cryptocurrency exchanges to steal wallet keys and transfer funds.Both advisories asked financial institutions to strengthen their security systems. The November 3 advisory recommends setting daily and per-transaction limits, checking beneficiary details, monitoring unusual transaction patterns, and ensuring that payments are made only after all required checks and approvals.The November 4 advisory calls for round-the-clock monitoring of cloud systems, stronger controls on access to payment APIs, separation of testing and production environments, regular security checks, and immediate action to isolate affected systems.Financial institutions have been asked to report cyber incidents to Cert-In within six hours of detecting them. Banks and NBFCs must report incidents to the Reserve Bank of India through its DAKSH portal within the same period.NBFCs and fintech companies have been advised to approach I4C for bulk reporting of fraudulent transactions so that the money can be frozen through the cyber fraud mitigation system.This comes two weeks after Cert-In issued an advisory warning of different categories of cyber incidents in which the attacker was an AI agent issuing commands, invoking tools and acting on the internet over extended periods.The incidents came to light through voluntary disclosures by AI model providers, OpenAI and Anthropic, a government official said.The agency said that the affected organisations could not detect the unauthorised access on their own. In some incidents, AI agents under research and evaluation exceeded their assigned task authorisation and made unauthorised access to Indian information and communication technology (ICT) assets.In others, cyber attackers deliberately used open-source agentic AI tools to conduct reconnaissance, exploitation, and financially motivated intrusions.