India's computer emergency response team has issued a critical warning: attackers are using AI-assisted tools against financial organizations and exploiting payment and disbursement APIs to initiate unauthorized fund transfers.
The CERT-In advisory, issued on November 4, 2026, covers incidents reported by non-bank financial companies, digital lending platforms, fintechs, payment service providers and virtual asset businesses. It says some organizations were targeted again after incomplete remediation.
For financial institutions, the warning raises a practical question: how quickly can a security team find an exposed route to a payment system, close it and verify that it is closed?
India: Attackers Are Targeting Payment APIs
CERT-In calls the activity CRAFTY FALCON. Its advisory describes the use of AI-driven offensive tools and reports attacks timed around holidays, weekends and other periods when monitoring may be reduced.
The reported attack patterns include internet-exposed user acceptance testing (UAT), staging or test environments, followed by movement toward production through shared credentials, tokens or network paths. CERT-In also describes attackers analyzing application code to develop tailored exploits and make unauthorized API calls.
An API may authenticate a request correctly and still need to check whether the caller is authorized to act on a specific merchant, account or transaction. That distinction matters when the interface can initiate a payment, change account details or interact with a disbursement workflow.
Financial Regulators Are Expanding Their Focus
In South Korea, the Financial Services Commission (FSC) reported 40 hacking incidents across 37 financial companies between September 30 and November 3, 2026. Approximately 473,000 information records were exposed, and smaller virtual asset, electronic finance and online lending businesses reported around KRW 1 billion in funds stolen from accounts they managed. The FSC had not identified individual users' fund losses at the time of its report.
The FSC's subsequent inspection plan covers external contact points, authentication and access controls, business and management APIs, and web development and test systems. In its October 14 update, the FSC said authorities were investigating whether incidents were linked and whether AI was used. The aggregate figures do not establish that all incidents formed a single AI campaign.
Japan's National Institute of Information and Communications Technology warned that attackers were analyzing publicly available Android applications and using the information to target their backend servers. Japan's Financial Services Agency and the Bank of Japan also requested short-term measures for financial institutions responding to frontier AI threats.
On November 12, the Australian Prudential Regulation Authority warned that AI could accelerate vulnerability discovery and exploitation, leaving financial institutions and important service providers less time to assess exposure and respond.
Recent Incidents Expose Different Parts of the Financial Supply Chain
The following incidents have different causes and impacts. They illustrate the range of systems that financial and digital-asset businesses depend on; they are not evidence of one coordinated campaign.
Binance: funds at scale. Binance reported approximately $6.34 billion in digital assets affected by its October 29 security incident. The FBI attributed the theft to North Korea. These disclosures do not establish that AI agents caused the incident.
Chainstack: development to production. The infrastructure provider reported that an attacker accessed its development environment and later parts of production, reaching customer account data and systems holding node credentials. Chainstack assessed with high confidence that the attacker used the ARTEX agentic framework. It asked customers to rotate credentials as a precaution.
GitBook: integration credentials. GitBook disclosed unauthorized access to infrastructure and characterized the attack as involving AI agents. Its incident response included credential rotation and guidance for customers with integrations and API tokens.
Chainalysis: security-service dependencies. Chainalysis reported exposure of customer information, password hashes and live tokens, and temporarily took some services offline during its response. The company said an external actor used offensive agentic techniques; its investigation updates distinguish this incident from the Binance theft.
Together, these cases point to three different security priorities: protect the systems that authorize and move funds; control credentials across development and production; and prepare for the loss of a critical provider. Each requires a different test and a different response owner.
What AI Changes for Financial Cybersecurity
AI tools can help attackers search for exposed assets and credentials, interpret responses, retry failed steps and coordinate activity across multiple targets. The result can be a shorter interval between exposure and impact.
Microsoft reported one case in which an AI agent progressed from searching GitHub for secrets to exfiltrating data from Azure Key Vault in approximately five minutes. It also observed misuse of valid application credentials for cloud discovery and data access. The report does not describe an unauthorized payment in that example.
Speed increases the importance of preparation. A team needs to know which credentials can reach payment systems, who can revoke them, what the affected service does and how the business will continue if a provider is unavailable.
What Payment Institutions Should Test Now
Start with an inventory of internet-facing applications, APIs, administrative consoles, staging environments and third-party connections. Map each asset to an owner and to the business functions it can reach.
Then validate the security boundaries around service identities and financial actions:
- Review where API tokens and keys are stored, who can use them, their permissions and how they are revoked.
- Separate test and production identities, secrets and network access.
- Verify object-level authorization for merchants, accounts, orders and recipients.
- Test the permitted state transitions for refunds, withdrawals and disbursements.
- Exercise credential compromise and third-party outage scenarios with security, engineering, operations and finance teams.
BlockSec's Perspective: Follow the Path to a Business Action
An external asset becomes more consequential when it can lead to a sensitive action. BlockSec looks at three connected areas: customer-facing applications and APIs; infrastructure and internal access; and suppliers and external dependencies. The assessment follows a reachable entry point to the data, operations or supporting systems it can actually reach.
For example, a test portal may be low impact if it is isolated and holds only synthetic data. Its risk changes if a shared credential can reach a production merchant API. An authorized assessment should establish whether that path exists and whether access, transaction-state and approval controls stop it at the intended boundary.
This approach helps prioritize remediation around three outcomes: unauthorized movement or incorrect crediting of funds, unauthorized access to data, and disruption to critical business services. An exposed system with a verified path to payment functions may need a different response from an isolated service with no sensitive access.
For digital-asset services, the scope may also include wallets, transaction construction, signing and settlement. BlockSec's crypto payment security playbook outlines how these components connect across a payment system.
How BlockSec Helps Financial Institutions Prepare
BlockSec's Payment Security Solution starts with an authorized external penetration test of the agreed assets. From an outside perspective, the team identifies reachable entry points, validates weaknesses, follows available paths and assesses their potential effect on funds, data and service availability. The initial black-box assessment does not require source code, internal accounts or special system access.
Findings document the entry point, conditions, evidence, business impact and remediation recommendations. The base engagement includes one retest after remediation. Depending on the findings and the institution's needs, further work can include architecture assessment, source-code audit, attack and incident-response exercises, or ongoing monitoring and response. Scope, safeguards and test conditions are agreed in writing before work begins. Learn more about BlockSec's payment security approach.
Assess your exposure to AI-accelerated attacks. Contact BlockSec at [email protected] to discuss the systems, testing boundaries and business workflows to include in your payment security assessment.
Series: Payment Security in the AI Era. The next article examines South Korea's financial-sector incidents and the regulatory response.



