Risk Management Strategies

SBI's Dark-Web Data Exposure: Will Your Cyber Policy Respond When Customer Data Leaks and You Cannot Prove Your Own Systems Were Breached?

SBI found 3,897 of its customer records on the dark web, and the source is still unknown. The case shows how cyber insuring clauses depend on where a leak started, and what lenders sharing KYC and lead data with vendors should fix before it happens to them.

Tarun Kumar Singh
Tarun Kumar SinghStrategic Risk & Compliance SpecialistAIII · CRICP · CIAFP
10 min read

Listen to this article

Audio version • 10 min read

cyber insurancedata breachvendor riskNBFCCERT-InDPDP

Last reviewed: November 2026

What SBI Actually Disclosed, and What It Did Not

On 21 October 2026, State Bank of India told the stock exchanges that it "has experienced an incident involving the loss of certain data". According to reporting by The Hacker News, the bank had found 3,897 customer records containing personal information inside a larger set of about 375,000 records offered on the dark web. The filing said the source of the data was not yet known.

The bank did not find the leak itself. The alert came on 16 October 2026 from CERT-In and the Defence Cyber Agency. SBI then engaged an independent CERT-In empanelled auditor and said there was no material impact on its operations.

The exposed fields reported were names, mobile numbers, dates of birth, gender, email and residential addresses, branch and CIF identifiers, lead IDs and loan-related metadata. Account numbers, passwords, card details, PAN and Aadhaar were not on the list. A threat actor using the name "Siriusim" claimed 375,000 CSV rows and 2,722 PDFs, including loan applications and KYC documents. Free Press Journal reviewed samples and could not establish whether the data came from SBI, a subsidiary or a third-party provider.

The presence of lead IDs and loan metadata is the detail worth noticing. Lead data typically moves between a lender, its sales channels, direct selling agents (DSAs), lead aggregators, CRM platforms and collection partners. Any of those could be the origin. Business Standard's 22 October round-up of Indian banking breaches, and Bank of Baroda's 27 July 2026 exchange disclosure of a cyber incident citing a potential business email compromise, show that these disclosures are now a recurring feature of the sector rather than one-off events.

Why the Source of a Leak Decides How the Policy Responds

Most Indian cyber insurance wordings are built around a defined trigger. The common versions read along these lines: a security failure or network security breach of the insured's computer system, or a privacy breach involving personal data in the insured's care, custody or control. The definitions of "computer system" and "care, custody or control" are where a leak of unknown origin gets argued.

Three scenarios produce three different coverage positions:

  1. The leak came from the insured's own systems. First-party breach-response costs and third-party privacy liability both respond in the ordinary way, subject to retention and sublimits.
  2. The leak came from a vendor holding data on the insured's behalf. Coverage depends on whether the policy extends "computer system" to systems operated by an outsourced service provider, and whether privacy liability responds to data held by others for the insured.
  3. The leak came from a party that held the data in its own right, such as an independent lead aggregator that sourced the same customers. The insured may have no loss trigger at all, even though its brand is attached to the leaked file.

The practical problem is that on day one nobody knows which scenario applies. Forensics may take weeks, and some vendor-side leaks are never conclusively traced. A policy that only responds once a breach of the insured's own network is proven leaves the insured funding the investigation that is supposed to establish coverage.

The care, custody or control test

When customer records sit with a DSA or a lead platform under a contract with the lender, many wordings still treat them as within the insured's control, because the lender is the party that collected and is responsible for the data. Some older wordings are narrower. Read the definition before you need it, not after a CERT-In alert arrives.

First-Party Breach-Response Costs While the Source Is Unconfirmed

The first costs of an incident like this arrive long before anyone knows where the data came from. Typical first-party heads under a cyber policy include:

  • Forensic investigation by an incident-response firm or CERT-In empanelled auditor
  • Legal advice on regulatory reporting and customer notification
  • Notification and call-centre costs if customers are informed
  • Credit and identity monitoring offered to affected customers
  • Public relations and crisis communications

The question is whether these costs are payable on a suspected breach or only an actual one. Better wordings trigger breach-response cover on a reasonable suspicion of a security or privacy breach, so that the forensic work that rules the insured's own systems in or out is itself covered. Weaker wordings require a confirmed event, which in a vendor-origin case may never be confirmed.

At renewal, ask the insurer to confirm in writing that forensic costs incurred to determine whether a breach occurred are covered even if the investigation concludes the source lay outside the insured's systems. Also check whether the insurer's panel vendors must be used, and whether a regulator-directed audit (for example, by a CERT-In empanelled auditor) counts as covered forensic cost.

Retention and waiting periods matter here too. Breach-response cover is usually subject to a retention that applies per claim. If the insurer later argues that the vendor leak and a related phishing campaign are separate events, the insured may pay two retentions. Clear related-matters or aggregation wording avoids that argument.

Privacy Liability When a DSA, Lead Aggregator or Vendor Leaked the Data

Third-party exposure follows the customer, not the vendor. If records containing names, mobile numbers, addresses and loan metadata are used for targeted fraud calls, the customer's complaint, regulatory grievance or civil claim will name the lender. Under the Digital Personal Data Protection Act, 2023, the data fiduciary remains responsible for personal data processed on its behalf by a data processor. Contracting the work out does not contract out the accountability.

That makes the privacy liability insuring clause, and its interaction with the vendor, central:

  • Does the clause respond to a privacy breach "by the insured or by any third party for whom the insured is legally responsible"? Without that phrase, a vendor leak may fall outside it.
  • Does it cover regulatory defence and, where insurable by law, penalties arising from a data protection proceeding? Insurability of fines remains a live question in India, so expect defence costs to be the more reliable part.
  • Does a contractual liability exclusion bite where the lender's exposure flows partly from its own data processing agreement with the vendor?

Subrogation against the vendor

Once an insurer pays, it will look to recover from the party that caused the loss. That right depends on the lender's vendor contract. A DSA agreement that caps liability at a month of commissions, or that waives claims against the vendor, can prejudice the insurer's subrogation rights, and some policies treat that waiver as a breach of condition. Check the vendor contract and the policy side by side. Our earlier analysis of third-party cyber and vendor risk covers the contract terms in more depth.

The Outsourced Service Provider Extension and Dependent Business Interruption

Many cyber policies now carry an outsourced service provider (OSP) extension, sometimes worded as "computer system includes systems operated by a service provider on behalf of the insured". This is the clause that converts a vendor-side leak from an argument into a claim. Its scope varies:

  1. Narrow: limited to named IT providers or hosting and cloud vendors, which may not include a DSA's laptop or a lead aggregator's CRM.
  2. Standard: any third party providing IT or data processing services under written contract, which usually captures CRM and lead-management platforms.
  3. Broad: any party holding personal data on the insured's behalf under contract, which captures field agents, DSAs and collection partners.

Lenders that run large DSA and connector networks should push for the broad form, or schedule the categories of partner explicitly. A sublimit is common, and a sublimit that is a small fraction of the main limit may be inadequate when vendor channels hold most of the lead data.

Dependent business interruption is a separate question

If a key vendor's system is taken offline after a breach, the lender may lose disbursement volume. That is a dependent or contingent business interruption claim, and it is often sublimited separately and subject to a longer waiting period. SBI said there was no material impact on operations, but a smaller NBFC that relies on one lead platform for most of its origination might not be so fortunate. The Korean lenders' loan-agent portal case, discussed in our post on cyber limits for NBFCs, shows how an agent-facing system can become the weak point.

CERT-In Reporting and SEBI LODR Disclosure as Policy Conditions

Regulatory reporting clocks start long before the coverage position is clear. The CERT-In Directions of 28 April 2022 require specified cyber incidents, including data breaches and data leaks, to be reported to CERT-In within 6 hours of noticing them or being notified of them. In SBI's case, CERT-In was itself the source of the alert, which is a reminder that the regulator may know before the insured does.

For listed entities, Regulation 30 of the SEBI (Listing Obligations and Disclosure Requirements) Regulations, 2015 requires disclosure of material events to the stock exchanges. SBI's 21 October filing came five days after the 16 October alert. Listed NBFCs and lenders need a pre-agreed view on when a dark-web sighting of customer data becomes disclosable, rather than deciding under time pressure.

These obligations interact with the policy in three ways:

  • Notice conditions. Most cyber policies require notice to the insurer "as soon as practicable". A regulator notification that precedes insurer notification by days can create an avoidable argument.
  • Consent to costs. Legal and forensic costs incurred without insurer consent may be challenged unless the policy carves out emergency or regulator-mandated costs.
  • Admissions. A public filing that attributes fault, or that states a vendor was responsible before forensics confirms it, can affect both the claim and any recovery.

A Checklist for NBFCs, Lenders and Anyone Sharing KYC or Lead Data

The SBI disclosure is a prompt for every business that shares customer data with partners to test its programme against a vendor-origin leak. Work through these points before renewal:

  1. Map the data. List every party that holds KYC documents, lead records or loan applications on your behalf: DSAs, connectors, lead aggregators, CRM vendors, collection agencies, document storage providers and subsidiaries.
  2. Read the trigger. Confirm that breach-response cover responds to a suspected breach, and that forensic costs to establish the source are covered whatever the result.
  3. Extend "computer system". Seek an outsourced service provider extension that captures every category from step 1, and check its sublimit against where your data actually sits.
  4. Fix privacy liability wording. Make sure it responds to breaches by parties for whom you are legally responsible, and that the contractual liability exclusion does not swallow DPDP-driven exposure.
  5. Align vendor contracts. Require partners to notify you of incidents within hours, to carry their own cyber or professional indemnity cover, and to preserve logs. Avoid waivers that prejudice subrogation.
  6. Plan for subsidiaries. If group companies share customer data, confirm that each is a named insured or that the policy's subsidiary definition includes them.
  7. Size the limit. A leak of a few thousand records may be modest. A leak of 375,000 rows and 2,722 KYC PDFs, if traced back to you, is a different scale. Our post on whether a 25 crore cyber limit is adequate works through the arithmetic.
  8. Rehearse the clocks. Run a tabletop exercise where the alert comes from CERT-In, the source is unknown and the exchange filing is due. Time how long it takes to notify the insurer.

Underwriters increasingly ask about third-party data governance at proposal stage. Evidence of a vendor inventory, contract clauses and access controls supports both terms and pricing.

What the SBI Case Means for Cyber Programme Design

The useful lesson from SBI's disclosure is not about any single bank. It is that a material share of customer data in Indian lending now lives outside the lender's own perimeter, and leaks are increasingly discovered by outside agencies rather than by the data owner. A cyber programme designed around "our network was hacked" will meet a growing number of incidents that do not fit that story.

Three design choices follow. First, buy breach-response cover that pays to find out what happened, not only after it is proven. Second, treat the outsourced service provider extension and privacy liability wording as core terms, not optional extras, because lead and KYC data flows through partners by design. Third, align vendor contracts with the policy so that the insurer can recover from a responsible partner and the insured has not quietly given that right away.

For lenders and NBFCs, the time to settle these points is before the next CERT-In alert, not during the five days between an alert and an exchange filing. Brokers placing cyber insurance for financial services clients should bring the vendor map to the renewal meeting and test the wording against a leak whose source is never confirmed.

About the Author

Tarun Kumar Singh

Tarun Kumar Singh

Strategic Risk & Compliance Specialist

  • AIII
  • CRICP
  • CIAFP
  • Board Advisor, Finexure Consulting
  • Developer of the Behavioural Underinsurance Risk Index (BURI)

Tarun Kumar Singh is a seasoned risk management and insurance professional based in Bengaluru. He serves as Board Advisor at Finexure Consulting, where he advises insurance, fintech, and regulated firms on governance, growth, and trust. His work spans insurance broker regulatory frameworks across India, UAE, and ASEAN, IRDAI compliance and Corporate Agency model reform, VC governance in insurtech, and MSME insurance gap analysis. He is the developer of the Behavioural Underinsurance Risk Index (BURI), a framework applying behavioural economics to underinsurance and insurance fraud risk.

Frequently Asked Questions

Will my cyber policy pay forensic costs if the leak turns out to have come from a vendor?
Only if the wording allows it. Better policies trigger breach-response cover on a reasonable suspicion of a breach, so forensic work that establishes the source is covered whatever the result. Narrower wordings require a confirmed breach of your own systems, which leaves you paying for the investigation if the source turns out to be a vendor or is never confirmed.
Are we liable to customers if our DSA or lead aggregator leaked their data?
Customers and regulators will usually look to the lender, because the lender collected the data and, under the Digital Personal Data Protection Act, 2023, remains responsible for processing done on its behalf. Your privacy liability cover should respond to breaches by parties for whom you are legally responsible, and your vendor contract should let you recover from the partner.
What is an outsourced service provider extension?
It extends the policy's definition of computer system to systems run by third parties on your behalf. Without it, a breach at a CRM vendor, DSA or lead platform may not trigger the policy at all. Check whether it is limited to named IT providers or covers any contracted party holding personal data, and whether its sublimit is adequate.
Does reporting to CERT-In count as notifying the insurer?
No. CERT-In reporting within 6 hours under the April 2022 Directions, and any SEBI LODR exchange disclosure for listed entities, are separate from the policy's notice condition. Notify your insurer through its breach hotline on the same day, and avoid public statements that attribute fault before forensics confirms it.

Related Glossary Terms

Related Insurance Types

Related Industries

Related Articles

Sarvada Intelligence

Ready to see Sarvada in action?

Explore the platform workflow or start a product conversation with our underwriting automation team.

Explore the platform