CERT-In Links Crafty Falcon to Backdoored ARTEX Docker Images Targeting 62 BFSI Consoles

CERT-In documented Crafty Falcon chaining public AI tools including ARTEX, PentAGI, and Cairn to run full autonomous attack sequences against banking APIs and payment gateways. Attackers first obtained hard-coded tokens and non-MFA endpoints, then used AI-generated scripts to bypass transaction limits and move funds via mule accounts. Post-exploitation included log deletion and endpoint tampering. Evidence consists of CERT-In's October and November 2026 alerts plus observed file-upload chains in PHP and Node.js applications. Procurement records and job postings show Indian financial regulators have accelerated AI-auditing mandates, yet the alerts reveal inconsistent enforcement of least-privilege boundaries on internet-facing consoles. Independent analysis of similar Docker supply-chain incidents indicates the 62 compromised instances likely stemmed from a single poisoned ARTEX image rather than multiple custom builds. The pattern matches earlier supply-chain compromises tracked in 2024-2025 where AI-assisted tooling accelerated script iteration from days to minutes. Official attribution names Crafty Falcon without releasing IOCs or confirming state ties, while technical artifacts point only to reuse of public offensive frameworks. Next steps include mandatory human-approval gates for high-risk AI agents under the new CERT-In guidelines; expect enforcement actions against non-compliant NBFCs within 90 days.