India's CERT-In Warns AI Hackers Are Stealing Wallet Keys From Crypto Exchanges

Published November 5, 2026, 12:45 PM PKT. Based on CERT-In Advisory CIAD-2026-0058, dated November 4, 2026 (primary source), and reporting by Cointelegraph and BeInCrypto on November 5, 2026 (secondary reporting).

India's national cybersecurity agency has warned that an AI-assisted hacking campaign is actively targeting cryptocurrency exchanges, stealing wallet keys and siphoning funds from wallets. The Indian Computer Emergency Response Team, known as CERT-In, named the threat actor CRAFTY FALCON in advisory CIAD-2026-0058, dated November 4, 2026, and rated the threat Critical, its highest severity level. The advisory says the campaign is active and ongoing, and that its finance incident response unit, CSIRT-Fin, continues to receive reports of successful attacks. The warning matters because it is not a hypothetical threat model: the advisory describes wallet keys already stolen and funds already drained, even though it names no affected exchange and gives no loss figure, so the crypto losses behind the warning are still undisclosed.

The November 4 advisory: what CERT-In found

The advisory is an update to CERT-In advisory CIAD-2026-0046, dated October 9, 2026, which first warned about threat actors exploiting payment and disbursement APIs for unauthorized fund transfers at financial firms. The November update says continued analysis of incidents reported by affected entities revealed a significant escalation of the campaign. The affected businesses now explicitly include cryptocurrency exchanges and virtual digital asset service providers, alongside non-banking financial companies, digital lending platforms, fintech entities, payment aggregators and payment service providers. CERT-In describes CRAFTY FALCON as an AI-assisted financial threat actor that surveys its targets with machine intelligence, operates at machine speed and strikes by invoking the victim's own payment and disbursement APIs. The agency's bluntest line is that the threat is active and ongoing, and it urges all banking, financial services and insurance entities to treat the campaign as an immediate and continuing threat.

Wallet keys stolen, funds siphoned

The crypto detail sits in the advisory's description of the attack pattern, in a single sentence: cryptocurrency exchanges and virtual digital asset service providers were targeted, with wallet keys stolen and funds siphoned from cryptocurrency wallets. That is the whole public record on the crypto side of the campaign. CERT-In does not name any exchange, does not say how many platforms were hit, and does not put a figure on the losses. Cointelegraph, which reported the warning on November 5, said it reached out to CERT-In and to Indian exchanges CoinDCX, CoinSwitch, WazirX, ZebPay and Mudrex for comment. The absence of names and numbers is itself a signal for readers: the advisory is written for defenders, not for the public, and its detailed indicators of compromise were shared only with financial firms through restricted advisories. What is public is the pattern, and the pattern is what exchanges are being told to defend against.

How the attacks work: the victim's own payment APIs

The advisory lays out the attack chain in unusual detail. The actor enters through internet-exposed non-production environments, such as user acceptance testing, staging or testing systems, then moves into production using shared credentials, tokens or network paths. Cloud infrastructure is abused to run commands and plant malicious files, with services such as AWS Systems Manager and Lambda used to blend the activity into legitimate cloud management channels. The actor reverse-engineers application code to build tailored exploits, and it defeats controls that assume only the legitimate client application will call an API by making unauthorized calls from altered or modified application packages. IP allow-listing is bypassed by sending traffic from whitelisted third-party fetcher IP pools, such as content-acceleration and app-store review fetchers, which any internet user can cause to fetch arbitrary URLs. Before moving large sums, the actor runs micro-value test transactions, such as one rupee, to validate that an unauthorized fund flow works. Newly published vulnerabilities are exploited as soon as they become public, including file-upload and deserialization flaws that lead to web shells in PHP and Node.js applications. Merchant settlement APIs of payment aggregators are also exploited or abused to move funds.

The AI toolkit: ARTEX, PentAGI, Cairn and CyberStrikeAI

The advisory names the AI tools it says were used against affected entities: the ARTEX family of large language model based tools, the PentAGI autonomous multi-agent AI penetration testing system, the Cairn attack search engine and the CyberStrikeAI framework. According to CERT-In, these tools execute complete attack chains autonomously at machine speed, generate and refine large numbers of tailored attack scripts in rapid succession, and adapt their payloads based on target responses. The supporting infrastructure rotates IP addresses within hours to days. The tools themselves are publicly available, which is why the advisory also warns firms to prohibit unauthorized security and attack tools on corporate infrastructure and to watch for AI coding-agent processes running with permission checks disabled. The AI angle is widening beyond this campaign: in October, cybersecurity firm CrowdStrike reported that an unidentified attacker used ARTEX alongside Claude Code and other AI models to breach South Korean financial firms and steal data, according to Cointelegraph. The same reporting noted that Binance has traced its $6.3 billion October hack to fake job interviews and a compromised AI tool, a separate incident that MABOnChain has covered in depth.

Attacks timed for holidays and low-monitoring windows

One detail in the advisory explains why it landed when it did. CERT-In says the same entities have been deliberately re-targeted, including after incomplete remediation, and that attacks have been timed to coincide with holidays, weekends and other low-monitoring windows to delay detection and response. The agency says successful attacks were reported even from non-banking financial companies and fintech entities that had already received the earlier restricted advisories and had undertaken remediation, which is why the update stresses assuming persistence after any incident, with full eradication, forensic verification and continued threat hunting. BeInCrypto noted that the Diwali festival falls on Sunday, November 8, a holiday weekend, though CERT-In does not mention Diwali in the advisory. The timing warning is the practical takeaway for users: the campaign is designed to strike when staffing is thinnest.

What exchanges are told to do now

For crypto platforms, the advisory's recommended actions are specific. Wallet keys should be stored in hardware security modules or cold custody, withdrawals should require multi-signature approvals, withdrawal patterns should be monitored for anomalies, and keys should be rotated immediately upon any suspected compromise. More broadly, the advisory tells firms to maintain heightened 24 by 7 security monitoring, with increased staffing and lowered alert thresholds during holidays, weekends and other low-coverage windows, and to extend real-time transaction monitoring to flag micro-value test transactions to new beneficiaries followed by high-value transactions, treating that pattern as a strong precursor of unauthorized fund transfer. It also tells firms to isolate testing and staging environments from production with no shared credentials, tokens, API keys, databases or direct network paths, and to treat client applications as untrusted by enforcing server-side validation of every transaction. Until affected exchanges disclose what was taken, the public picture stays incomplete: the advisory proves the campaign reached crypto wallets, but not how much it cost or whose keys were taken.

Sources

• CERT-In, Advisory CIAD-2026-0058: Escalation of AI-Driven Attacks on BFSI Entities by Threat Actor "CRAFTY FALCON" (2026-11-04), CERT-In advisory CIAD-2026-0058, November 4, 2026 (primary source: severity Critical, update to CIAD-2026-0046 dated October 9, 2026, CRAFTY FALCON designation, crypto exchanges targeted with wallet keys stolen and funds siphoned, attack pattern and recommended actions)
• Cointelegraph, India Warns AI Hackers Are Targeting Crypto Exchanges, November 5, 2026 (secondary reporting: advisory summary, outreach to CERT-In, CoinDCX, CoinSwitch, WazirX, ZebPay and Mudrex, CrowdStrike October report on ARTEX use against South Korean financial firms)
• BeInCrypto, CERT-In Warns AI-Powered Hacker Group Is Hitting India's Crypto and Fintech Firms, November 5, 2026 (secondary reporting: November 4 warning rated Critical, AI tools PentAGI and CyberStrikeAI, WazirX July 2024 theft of more than $230 million as context, Diwali on November 8)

This article is news reporting and is not investment advice.