Banking agentic AI attack — Nov 2026

The Indian Computer Emergency Response Team (CERT-In) has identified an active campaign by a threat actor group known as Crafty Falcon, which is leveraging AI-driven offensive security tools to target the Banking, Financial Services, and Insurance (BFSI) sectors. These autonomous agents are being used to execute complex attack chains at machine speed, enabling the rapid theft of funds and sensitive credentials from financial entities.

What happened

According to CERT-In, threat actors are utilizing tools such as ARTEX, PentAGI, Cairn, and CyberStrikeAI to autonomously adapt payloads and generate tailored attack scripts. These agents exploit internet-facing applications, exposed APIs, and privileged endpoints to bypass application-level controls, including KYC, beneficiary verification, and transaction limits. By invoking fund-transfer APIs directly from within an entity's allow-listed infrastructure, attackers have successfully moved funds to mule accounts.

Beyond direct financial theft, the agents are used to reverse-engineer application codebases, deploy malicious payloads, and pivot from non-production environments into production systems. The attackers also employ techniques to delete logs and tamper with endpoint security tools to evade detection.

Separately, the security of these AI tools themselves has been compromised. Researchers identified backdoored versions of the ARTEX penetration testing tool hosted on Docker Hub. One specific malicious image, "tomicdd/artex," contains a hidden function that grants an unauthenticated root shell to attackers. As of early November 2026, 62 instances of these backdoored consoles were identified globally, exposing the infrastructure of users who deployed the unvetted tools. CERT-In has responded by publishing new Artificial Intelligence Security Auditing Guidelines, which emphasize the need for human oversight, permission boundaries, and the principle of least privilege for all AI systems operating within critical infrastructure.