Home / Blog / An AI Debt Collection Service Exposed 6.6 Million Accounts…
Tech News

An AI Debt Collection Service Exposed 6.6 Million Accounts and 315,000

Two security flaws in CreditNirvana exposed 6.6 million borrower accounts and 315,000 call recordings across prominent Indian financial services firms.

By Dillip Chowdary • Nov 19, 2026 • Source: Wired

An AI Debt Collection Service Exposed 6.6 Million Accounts and 315,000
Image: Wired

At a glance

Product
CreditNirvana
Parent Company
Perfios
Accounts Exposed
6.6 million accounts
Call Recordings
315,000 recordings
Disclosure Timeline
Patched in June, disclosed October
Reporting Agency
CERT-In

Two vulnerabilities in the Indian debt collection platform CreditNirvana exposed 6.6 million borrower accounts and 315,000 call recordings from its agentic contact center. Owned by Indian fintech firm Perfios, the platform reads borrower context to autonomously execute recovery actions within set policies. Details of the system exposure were reported in Wired's report following an independent security investigation.

This breakdown details the two vulnerabilities, the exposure of borrower records, and the remediation coordinated with national cybersecurity authorities. It provides technical analysis for engineering teams and compliance officers deploying autonomous software across regulated financial environments.

01Coordinated disclosure and rapid vendor remediation

Security researcher Eaton Zveare identified two flaws in CreditNirvana's infrastructure and reported them directly to the Computer Emergency Response Team of India, known as CERT-In. Following CERT-In notifications, the engineering team patched the production systems within a single day in June. Public disclosure occurred in October after remediation was verified.

The technical research showed that the vulnerabilities exposed sensitive personal data belonging to millions of borrowers across India. Because the service handles debt recovery, the repositories held detailed documentation on unpaid accounts and credit histories. Zveare noted that while maintainers fixed the flaws rapidly, the vulnerabilities were potentially present in code for years.

Note: CreditNirvana and parent company Perfios did not return requests for comment regarding the system vulnerabilities. Neither entity has clarified the duration of the data exposure or the total number of borrower records that were accessed by external parties.

02Agentic call center exposure and campaign control

The high severity vulnerability exposed CreditNirvana's call center infrastructure, leaving approximately 315,000 call recordings accessible without authorization. These audio files contained direct conversations between automated agents, recovery staff, and borrowers regarding unpaid debts. The system also exposed post-call recommendations generated by the platform for subsequent recovery steps.

Beyond passive data exposure, the flaw permitted direct manipulation of the collection system. An attacker could exploit the bug to manage autonomous software agents and orchestrate unauthorized calling campaigns through CreditNirvana's agentic center. This capability created severe risks, allowing unauthorized actors to redirect automated communications sent to borrowers.

Advertisement

Tech Pulse Daily

Get tomorrow's pulse first

Join engineers who read Tech Pulse before stand-up. Free, weekday mornings.

03Platform vulnerability ratings and system impact

CERT-In evaluated the security disclosures and classified the two flaws under high and critical severity designations. The technical assessment separated the exposure of autonomous call records from the broader compromise of the core debt collections platform used by financial institutions.

VulnerabilitySeverityImpacted SystemPrimary System Consequences
Call Center FlawHighAgentic calling serviceExposed 315,000 recordings and allowed unauthorized management of autonomous campaigns
Main Platform FlawCriticalCore collections portalExposed 6.6 million accounts, identity documents, session impersonation, and agent tracking

The critical vulnerability directly exposed confidential customer records across three prominent Indian financial services firms using CreditNirvana for debt recovery. By compromising the centralized system, attackers accessed consolidated borrower records across all three corporate clients. Zveare emphasized that major Indian fintech companies rely on CreditNirvana to autonomously recover outstanding loans.

04Administrative session impersonation and borrower document exposure

Exploiting the critical flaw allowed attackers to bypass authentication controls on the primary platform. Technical findings from CERT-In confirmed that attackers could enumerate platform users, establish sessions as arbitrary users including administrators, and modify sensitive records. Impersonation granted full operational privileges across the underlying system databases.

The compromised repositories contained extensive documentation regarding 6.6 million individual borrower accounts. Exposed files included official identification cards, credit reports, court records, legal filings, case notes, and millions of emails and text messages regarding unpaid balances. The flaw also allowed attackers to track real-time locations of human collection agents in the field.

05Autonomous data synthesis and operational exposure

The security findings demonstrate significant operational risks in centralized algorithmic services that consolidate sensitive data to automate recovery decisions. Zveare pointed out that as automated systems synthesize data across disparate financial records, single points of failure create extensive systemic exposures.

CreditNirvana designed its architecture to ingest complete contextual profiles for each borrower, synthesizing information from court cases, credit bureaus, and direct communications. While the one-day patch resolved the immediate vulnerability endpoints, the incident stands as an important case study in securing autonomous artificial intelligence systems handling sensitive financial data.

Frequently asked questions

What data was exposed in the CreditNirvana security incident?

The vulnerabilities exposed 6.6 million borrower accounts containing credit reports, identification cards, court records, legal files, emails, texts, and call transcripts. Additionally, around 315,000 call recordings from the autonomous call center were left accessible.

Who discovered the vulnerabilities in CreditNirvana?

Independent security researcher Eaton Zveare discovered the two flaws and coordinated their remediation with India's Computer Emergency Response Team (CERT-In).

How quickly did CreditNirvana resolve the vulnerabilities?

CreditNirvana fixed the security flaws within one day in June after notification, followed by public disclosure in October.

What administrative risks did the critical vulnerability introduce?

According to CERT-In, attackers could enumerate users, establish sessions as arbitrary or administrative users, modify sensitive information, and track the real-time location of human debt collection agents.

Sources

Dillip Chowdary

Author

Dillip Chowdary

Writes Tech Bytes coverage of AI, engineering, and the tools that actually ship. Editor of Tech Pulse Daily.

Related on Tech Bytes

Advertisement

5-min tech signal

Weekday briefing for engineers who skip the noise.

No spam · Unsubscribe anytime

Advertisement

✈️ CareerPilot

Your AI job-search copilot

Match your resume against live Ashby, Greenhouse & Lever openings — fit scores, job-specific resume optimization and email alerts.

Find matching jobs →

Free Tools

Browse all tools →