India's cybersecurity watchdog, the Indian Computer Emergency Response Team (Cert-In), has warned banks, non-banking financial companies (NBFCs) and fintech firms about an escalating AI-driven cyberattack campaign linked to unauthorised fund transfers. The watchdog has asked financial institutions to strengthen their security systems and remain alert to potential breaches.

In an advisory issued on Wednesday, Cert-In identified the campaign as "Crafty Falcon". It uses publicly available artificial intelligence (AI) tools to automate cyberattacks and execute them at high speed. Attackers use these tools to generate and modify large numbers of customised attack scripts.

The latest advisory updates an earlier warning issued on October 9 about threat actors exploiting payment and disbursement APIs to carry out unauthorised transfers.

Cert-In said its analysis of incidents reported by affected organisations showed a significant rise in the campaign. "The threat is active and ongoing. Computer Security Incident Response Team-Finance (CSIRT-Fin) continues to receive reports of successful attacks, including from NBFCs and fintech entities that had already received the earlier restricted advisories and had undertaken remediation," the advisory said.

Some organisations have faced repeated attacks despite receiving earlier warnings. Attackers have also timed their operations around weekends and holidays, when security monitoring is often reduced, delaying detection and response.

Cert-In identified several publicly available offensive security tools, including the ARTEX family of large language model (LLM)-based tools, PentAGI, an autonomous multi-agent AI penetration testing system, the Cairn attack search engine and the CyberStrikeAI framework.

These tools can independently execute attack chains, create customised scripts and modify their methods based on how target systems respond.

Attackers are exploiting internet-facing testing and staging environments to enter production systems through shared credentials, tokens and network connections. They are also abusing cloud services such as AWS Systems Manager and Lambda to execute commands and deploy malicious files while disguising their activities as legitimate operations.

Other methods include reverse-engineering application code, modifying application packages to make unauthorised API calls and bypassing IP-based security controls through whitelisted third-party services.

Attackers have also used micro-value transactions, sometimes as low as ₹1, to test unauthorised transfers before moving larger amounts. Payment aggregator merchant settlement APIs have also been exploited. Cert-In has flagged file-upload and deserialisation vulnerabilities in PHP and Node.js applications. Cryptocurrency exchanges and virtual digital asset service providers have faced attacks involving wallet key theft and fund transfers.

Cert-In has asked financial institutions to strengthen round-the-clock monitoring, increase security staff during weekends and holidays, and monitor cloud audit logs in real time. It has also recommended isolating testing environments from production systems, avoiding shared credentials and API keys, and restricting cloud management services and internet access for serverless functions.

The watchdog has called for server-side transaction validation, API request signing using nonces and timestamps, and app attestation to identify modified applications. It has also advised faster security patches for internet-facing applications and identity-based safeguards instead of IP-based controls for third-party services.

Banks and fintech firms should flag small transfers to new beneficiaries followed by larger transactions. Cryptocurrency platforms should protect wallet keys using hardware security modules or cold custody, require multi-signature withdrawal approvals and monitor unusual withdrawals.

Cert-In has also recommended restricting unauthorised AI security tools, monitoring LLM API usage and adopting zero-trust security, micro-segmentation and identity-based access controls.

Organisations facing suspected breaches should isolate affected systems, preserve forensic evidence and logs, rotate credentials and report incidents to Cert-In or CSIRT-Fin.