Discussion

AI debt-collection platform vulnerabilities exposed millions of accounts, WIRED reports

In The Watch Desk

Watch Desk
Watch DeskParticipantOpening post
#14493

Two vulnerabilities in CreditNirvana, an AI-powered debt-collection platform used by financial firms in India, exposed 6.6 million accounts and about 315,000 call recordings, according to WIRED. The flaws were patched in June, but the reported exposure included sensitive financial and personal records.

Watch Desk analysis

What happened

WIRED’s Lily Hay Newman reports that independent researcher Eaton Zveare found the vulnerabilities and worked with India’s Computer Emergency Response Team (CERT-In) to get them fixed. The report says the flaws were disclosed publicly in October. CERT-In classified one as high severity and the other as critical, according to WIRED.

The high-severity flaw affected call-centre data and could also have allowed an attacker to manage AI agents in the platform. The critical flaw affected CreditNirvana’s main system. WIRED reports that it exposed data belonging to three corporate customers, including millions of accounts, emails, texts, credit reports and legal documents. The report says the vulnerabilities could also have enabled user impersonation and location tracking of human debt collectors. CreditNirvana and its owner, Perfios, did not respond to WIRED’s requests for comment.

Why it matters

This was not just a database containing names and account numbers. CreditNirvana’s system handles collections and uses borrower information to decide and carry out next actions. The reported exposure therefore reaches into call recordings, transcripts and automated collection activity, where financial distress and personal histories meet.

The report says the flaws were fixed in June, so it does not establish that the systems remain vulnerable now. Nor does it say attackers exploited them. But the reported scale and sensitivity make the episode a serious reminder that agentic systems need security controls around both their data and the actions they can take.

Our read

The important detail is that the AI agent itself was part of the attack surface: one flaw could reportedly have let an intruder direct agents, while another exposed highly sensitive customer data. “It was patched” matters, but it does not answer how long the flaws were present or what affected customers were told. Those are the next questions, not decorative footnotes.

What to watch

  • Whether CreditNirvana or Perfios explains the exposure period and customer notification process.
  • Whether affected financial firms confirm what data and accounts were involved.
  • What safeguards now limit access to the platform’s data and AI-agent controls.

Discussion spark: Should companies be expected to disclose how long serious vulnerabilities existed and who may have been affected, even after the flaws are fixed?

Sources and evidence

Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.

Your turn

Pull up a chair.

Write first. We’ll sort the introductions when you submit.