SBI data breach

What the SBI Data Exposure Tells Indian Banks About DPDP Readiness

Picture of Shikha Patra
Shikha Patra
Product Marketing Manager
• ⏱︎ 8 min read


TL;DR

  • What happened: On 21 October 2026, State Bank of India told the stock exchanges that a preliminary review of roughly 375,000 records found on the dark web identified 3,897 customer records containing personal information.
  • What was exposed: Names, mobile numbers, dates of birth, email and residential addresses, plus branch and CIF identifiers and metadata tied to loan and lead records. No passwords, PINs or OTPs were listed.
  • What is still unknown: The source, the entry point and the full scope. An independent CERT-In-empanelled agency is investigating.
  • Why it matters for DPDP: From 13 May 2027, a bank in this position must tell the Data Protection Board and every affected customer, then file a detailed report within 72 hours. Today, reporting runs mostly to regulators, and customers may never hear directly.
  • What to do now: Know which APIs carry personal data, who calls them and how much they read. Without that, a 72-hour report with an accurate count is very hard to produce.

What happened

India’s largest bank confirmed a data loss incident in a 21 October 2026 filing with the BSE and NSE. The alert came five days earlier, on 16 October, through SBI’s threat intelligence channels, which its filing names alongside the Defence Cyber Agency, and from CERT-In. Both flagged suspected customer data on the dark web.

The same day, monitoring account CyberPulse56 posted on X about a claim of roughly 3.75 lakh CSV entries and 2,722 PDF files, which reports linked to loan applications and account-opening documents. SBI’s own review of those roughly 375,000 records found 3,897 customer records carrying personally identifiable information.

SBI activated its incident response and containment procedures and engaged an independent agency empanelled by CERT-In. That investigation will test whether the data is authentic and complete, where it came from and how far the exposure goes. The bank says core operations, financial performance and business continuity are unaffected.


What we know and what we don’t

The disclosure confirms data loss. It does not confirm a breach of SBI’s core banking systems, and it names no entry point, vulnerability, malware or compromised server. Anyone writing about this incident, including us, should hold that line until the forensic report lands.

Confirmed by SBIStill open
3,897 customer records with PII among about 375,000 reviewedWhether the data is authentic and complete
Fields include contact, identity, branch, CIF and loan-linked metadataWhere it came from: SBI, a vendor, a partner or a channel
No passwords, PINs or OTPs listed; no stolen funds reportedHow it was accessed and over what period
Independent CERT-In-empanelled agency engagedWhether affected customers will be told directly

One caution on the numbers: 375,000 rows is not 375,000 victims. Record counts include duplicates, and SBI’s own count of records with PII is 3,897.

Why 3,897 records still matter

A small count can carry a large fraud risk when the fields are the right ones. This set goes well beyond a contact list.

  • Identity plus contact: Name, date of birth, mobile number, email and home address are enough to pass many caller-verification checks.
  • Bank context: Branch details and CIF identifiers let a fraudster sound like the customer’s own branch.
  • Loan and lead metadata: References to a live loan or a recent application give a scam call a believable reason to exist, such as an EMI problem, a disbursal step or a KYC update.

None of this needs a password to cause harm. It feeds targeted phishing and vishing, which CERT-In has repeatedly warned about in the form of fake bank messages and urgent account alerts. Unlike a password, a date of birth or home address cannot be reset.

The Internet Freedom Foundation has pointed to the other half of the problem. Under current rules, the affected customers may not be told directly, so they cannot be on guard for exactly the calls this data enables.

The DPDP lens: what changes from May 2027

If this incident happened after 13 May 2027, SBI would owe the Data Protection Board a detailed report within 72 hours and would have to inform each affected customer. The DPDP Rules were notified on 14 November 2025, and the core duties, security safeguards under Rule 6 and breach intimation under Rule 7, apply from 13 May 2027. A proposal to bring that date forward is under discussion.

ObligationTodayFrom 13 May 2027 under DPDP
Report to regulatorsCERT-In within 6 hours; RBI under its cyber frameworkSame, plus intimation to the Data Protection Board without delay
Tell affected customersNot directly required by CERT-In or RBI rulesRequired, under Section 8(6) of the Act and Rule 7
Detailed breach reportNo DPDP equivalentTo the Board within 72 hours: nature, extent, timing, likely impact, mitigation
Access logs for personal dataCERT-In: ICT logs for 180 days, kept in IndiaRule 6: visibility of who accessed personal data, logs kept for one year
Penalty for failed safeguardsSector penaltiesUp to ₹250 crore

The 72-hour report is where most programs will struggle. It asks for scope: which data, which customers, through which system, and when. SBI’s own timeline shows the difficulty. Five days after the first alert, the bank could confirm a record count but not the source. That is a normal pace for forensics. Under DPDP, it is past the deadline.

Rule 6 also changes what counts as reasonable security. It names specific safeguards: masking or tokenizing personal data, controlling access to the systems that hold it, keeping visibility of who accessed it through logs and monitoring, and showing that these controls work in practice. Data Fiduciaries stay accountable when a processor or partner handles the data on their behalf.

Where personal data actually leaves: the API layer

In Indian banks and fintechs, core systems are usually well guarded. Personal data tends to leave through the paths built to share it: mobile app endpoints, loan origination and lead management integrations, partner and DSA portals, and AI agents holding broad tokens. Every one of those paths is an API.

We don’t know yet whether SBI’s data moved through any of these. But the fields in the set (loan references, lead IDs and branch metadata) are the kind that flow between banks and their partners every day. That is reason enough to ask where your own lead and loan data travels.

India already has a clear example of how this goes wrong. In September 2025, researchers found that any logged-in user of the income tax e-filing portal could read another taxpayer’s records by swapping their own PAN for someone else’s in the portal’s API requests. The responses carried names, addresses, dates of birth, bank details and Aadhaar numbers, on a portal with more than 135 million registered users. CERT-In acknowledged it on 30 September, and it was fixed by 2 October. It is a textbook case of broken object-level authorization (BOLA), the first risk in the OWASP API Top 10. The cost of BOLA and BFLA flaws goes well beyond the fix.

That case maps cleanly to what DPDP Rule 6 asks for:

Data security posture management (DSPM) and data loss prevention (DLP) tools show where personal data sits and where employees send it. Neither shows who is reading it through each API, and whether they should be. A WAF or API gateway sees each request, but cannot tell whether the caller should own the record it returns. That takes business logic security, built on a business logic graph of how users, objects and functions relate. Regulators have noticed: SEBI’s CSCRF and RBI’s direction for payment system operators now name API authentication, authorization and threat protection directly. For payment operators, we break these down into seven API security requirements for payment transactions.

Five questions every Indian bank should answer before May 2027

If any answer is “we’d have to check,” that is the gap to close first. Our API security buyer’s checklist covers how to evaluate the tools that close it.

  1. Inventory: Do we have a live list of every API that returns personal data, including shadow, dormant and partner-facing ones?
  2. Callers: For each of those APIs, do we know which users, partners, processors and AI agents call it, and how much each one reads?
  3. Authorization: Do our release tests prove that one customer, or one partner, cannot read another’s records?
  4. Anomalies: If a partner account read ten times its normal volume over a week, from new IPs, would we see it before it showed up on the dark web?
  5. Reporting: Could we send the Data Protection Board a detailed report within 72 hours, with a count of affected records and the path the data took?

Get DPDP-ready at the API layer

Our whitepaper, DPDP-Ready Data Protection at the API Layer, maps DPDP Rules 6, 7 and 13, CERT-In, SEBI CSCRF and RBI expectations to specific API-layer controls, with a one-page India requirement map for your compliance team and auditors.

Download the DPDP whitepaper and book a demo to see how AppSentinels gets your bank DPDP-ready at the API layer.

Related Content