If you run payment or disbursement APIs at a bank, NBFC, insurer or fintech, assume an automated attacker may already be probing them, and that a past fix may not be enough. CERT-In has told the sector to treat a new AI-driven campaign as an immediate and continuing threat. The weak spots are the money-moving interfaces you built yourself, and the quiet hours when few people are watching.
What happened
According to Inc42, CERT-In issued an advisory on 4 November 2026 naming a threat actor it calls CRAFTY FALCON. CERT-In describes it as AI-assisted: it uses machine intelligence to pick targets, attacks at machine speed and abuses victims’ own payment and disbursement APIs to start unauthorised fund transfers. Its finance-focused team, CSIRT-Fin, reports successful attacks continuing across the ecosystem.
A key concern in the advisory is that the actor kept targeting entities that had received earlier restricted advisories and had undertaken remediation. CERT-In says some were deliberately targeted again, including cases where remediation was incomplete. Attacks were also timed for holidays, weekends and other periods of reduced monitoring. The campaign uses publicly available offensive tools, including the ARTEX family of LLM-based tools, the PentAGI multi-agent penetration testing system, the Cairn attack search engine and the CyberStrikeAI framework. Per CERT-In, these can run full attack chains on their own and adapt payloads to how a target responds. Attack infrastructure reportedly rotates IP addresses within hours to days.
The Hacker News, in related coverage, also describes the activity as active and ongoing and says it involves publicly available AI-driven offensive tools. Its headline refers to a backdoored ARTEX item exposing 62 consoles to root access, but we have only the headline for that claim and cannot confirm the details.
Why it matters
This advisory follows a separate critical CERT-In warning from October, which covered attackers using public agentic AI tools for financially motivated intrusions into Indian organisations. Inc42 notes that autonomous agents ostensibly deployed for research had accessed live systems without authorisation and read records with personal data. It adds that the Indian Cyber Crime Coordination Centre has warned of cases where malicious AI agents used compromised financial systems to move amounts running into crores of rupees.
The pattern suggests the cost of attacking has dropped. Tools anyone can download now chain reconnaissance, script writing and exploitation, so a small group could test many firms at once. Fixes that rely on a one-time patch or a static blocklist look fragile against infrastructure that changes within hours. The return to previously warned firms also suggests attackers may check whether remediation was finished, not just started.
What it means for you
Start with the payment and disbursement APIs. List every endpoint that can initiate a transfer, who or what can call it, and which credentials guard it. Check transaction limits, step-up approvals for unusual payouts and velocity controls that do not depend on a human noticing.
Reopen past remediation. If you received a restricted advisory, confirm each action is closed and verified, not just ticketed. CERT-In says some re-targeted organisations had incomplete remediation, so partial fixes carry real risk.
Close the monitoring gap. Staff alerting and on-call cover for weekends and holidays, and make sure someone can freeze disbursements outside office hours. Ask your SOC or managed provider how fast they can ingest the detection rules CERT-In shares and how they handle rotating IPs. If you are not on the restricted advisory list, ask CERT-In or your regulator contact how to get on it.
Question vendors too. Ask fintech partners and API providers how they rate-limit and anomaly-detect automated callers. Treat AI testing tools as risky in your own estate: restrict where autonomous agents can run and what they can reach. Startups should budget for this early; our startups coverage tracks how security spend is becoming a diligence item.
What to watch
- Further CERT-In or I4C advisories, especially any that add indicators or name new tools.
- Whether regulators turn the guidance into formal API security or after-hours monitoring requirements.
- Public disclosures of losses at financial firms, which would show the scale of the campaign.
- More detail on the ARTEX-related exposure mentioned in The Hacker News headline.
