Skip to content
Brief Security & Trust ·

Flaws in Indian AI debt collector exposed 6.6 million accounts

Researcher Eaton Zveare says two bugs in CreditNirvana opened borrower data and call recordings; both are now fixed.

In brief
  • Two flaws in CreditNirvana, owned by Perfios, exposed 6.6 million accounts and about 315,000 call recordings. Perfios fixed both in June, the researcher said.
  • One flaw was a login check the server did not verify. The other was a secret key left in public web page code.

Independent researcher Eaton Zveare said on 18 November that two flaws in CreditNirvana exposed 6.6 million accounts. CreditNirvana is an Indian debt collection platform that uses AI agents, and Perfios owns it. The accounts belonged to customers of three firms: Aditya Birla Capital Finance, BharatPe and International Asset Reconstruction Company. Zveare also said a separate AI call centre system gave access to about 315,000 call recordings from Aditya Birla Capital Finance customers. Wired reported the same findings.

Zveare said Perfios fixed both flaws on 2 June, the day India's cyber agency, CERT-In, confirmed it would act. CERT-In issued its public note on 26 October. Wired reported that CERT-In rated one flaw critical and the other high. Wired said CreditNirvana and Perfios did not answer its requests for comment. The sources do not say whether anyone other than Zveare used the flaws. Wired said Zveare believes the bugs may have been there for years, but no start date is given.

Zveare said the main platform did not check the login token it was sent and trusted a user name in the request instead. The call centre system had a secret key sitting in public web page code. Teams that buy AI tools to handle customer data can ask vendors two plain questions: how the server checks logins, and where secret keys are kept.

A WebPulse Brief: a short report of an important event, written by the WebPulse Newsroom with AI assistance and checked against the reporting below. When there is more to explain, we follow up with a full story. How we use AI.

Reporting: Eaton Zveare, independent security researcher (eaton-works.com).