AI Security · Privacy
AI debt collection platform exposed millions of sensitive records
CreditNirvana, an AI-driven debt collection service, exposed 6.6 million accounts and 315,000 call recordings due to two serious vulnerabilities. The flaws were fixed in June, but sensitive data was at risk for years.
Ross Grant2 min read

The short version
A researcher found two security flaws in CreditNirvana, an AI debt collection service. The flaws exposed millions of people's personal and financial data, including call recordings and credit reports. The company fixed the issues quickly after being told. Sensitive information was at risk for years before the problems were found.
By the numbers
- 6.6 millionaccounts exposed by the vulnerabilitiesWired
- 315,000call recordings exposedWired
What happened?
CreditNirvana, an AI-powered debt collection platform, exposed 6.6 million accounts and 315,000 call recordings due to two serious vulnerabilities. The flaws were fixed in June 2026 after a researcher reported them to India's CERT-In. Sensitive data, including credit reports and legal documents, was at risk while the bugs were active. Only Wired has reported this so far.
What was taken
According to Wired, two vulnerabilities in CreditNirvana, an AI-powered debt collection platform owned by Perfios, exposed a large trove of sensitive information. The exposed data included 6.6 million individual accounts and about 315,000 call recordings from the company's automated call centre. The records contained credit reports, ID cards, case management details, legal documents, and communications such as emails, texts, and call transcripts. Some of the data related to unpaid accounts and debt collection cases, making it especially sensitive.
How the flaw works
Independent researcher Eaton Zveare discovered and reported the vulnerabilities. One flaw, rated "high" severity by India's Computer Emergency Response Team (CERT-In), exposed call centre data and could have let an attacker manage AI agents or launch their own automated campaigns. The other, rated "critical" severity, affected the main platform and could have allowed an attacker to impersonate any user, including administrators. This would let them access or change sensitive information, track the location of human debt collectors, and enumerate all platform users. CERT-In confirmed that successful exploitation could have given an attacker broad access and control over the system.
The response
The vulnerabilities were patched in June after Zveare worked with CERT-In to alert the company. The issues were publicly disclosed in October. According to Zveare, CreditNirvana fixed the bugs within a day of being notified. However, he noted that the flaws may have existed for years before being discovered. Neither CreditNirvana nor Perfios responded to Wired's requests for comment.
Who is affected
The main vulnerability exposed data belonging to three major Indian financial services firms that use CreditNirvana for debt collection. Millions of individuals whose accounts were managed by these firms were also affected. The incident highlights the risks when AI-driven platforms process and store large volumes of personal and financial data.
What you should do
While the vulnerabilities have been fixed, the incident is a reminder for organisations using AI platforms to review their security controls and ensure regular vulnerability assessments. Individuals whose data may have been exposed should be alert for signs of identity theft or phishing attempts, especially if they have had debt collection dealings with Indian financial firms.
Why it matters
AI platforms are handling more personal and financial data than ever. This case shows how a single flaw can put millions of people’s sensitive information at risk. It also highlights the need for strong security checks in AI-driven services.
Key points
- 6.6 million accounts and 315,000 call recordings were exposed.
- The flaws allowed attackers to impersonate users and access sensitive data.
- Vulnerabilities were fixed quickly after being reported, but had existed for years.
- AI-powered platforms can create new risks for personal and financial data.
- Regular security reviews are essential for organisations using AI services.
