AWS Consulting ·
RBI Cybersecurity Directions 2026 on AWS: a clause-by-clause control map for commercial banks
The RBI’s 2026 Cybersecurity Directions for commercial banks took effect on 31 July 2026. Here is how the clauses that touch AWS workloads map to AWS services and settings, and the evidence a bank should keep for each.
By Upvantix team
Under the RBI’s Cybersecurity Directions, 2026 for commercial banks, in force since 31 July 2026, AWS workloads need these day-to-day controls: audit logging and log monitoring (paragraphs 93–97), patch and change management (98–103), access control and multi-factor authentication for privileged users (104–113), VA at least every six months and PT at least every 12 months for critical and customer-facing systems, with an approach that also covers cloud-hosted systems (149–161), half-yearly DR drills and tested backups (163–174), incident reporting on RBI’s DAKSH platform within six hours of detection (182), and a Cyber Security Operations Centre with SIEM (143, 212–223).
The Reserve Bank of India (Commercial Banks – Cybersecurity, Technology: Risk, Resilience and Assurance Framework) Directions, 2026 (RBI/DoS/2026-27/410, dated 31 July 2026) took effect on issue and apply to commercial banks: banking companies other than small finance banks, payments banks and local area banks, plus corresponding new banks and the State Bank of India. They replace the earlier cybersecurity and IT governance instructions for these banks. The bank stays accountable for outsourced and cloud arrangements. Where a bank’s cloud use is material IT outsourcing, the RBI’s Managing Risks in Outsourcing Directions, 2025 add cloud-specific duties.
AWS gives banks the building blocks — CloudTrail, CloudWatch Logs, AWS Config, Security Hub CSPM, GuardDuty, Amazon Inspector, IAM Identity Center, AWS KMS and CloudHSM, AWS Backup — but the evidence, the drills and the reporting are the bank’s to produce.
This page is general information, not legal or regulatory advice. Confirm your obligations with your compliance and legal teams and the current text on the RBI website.
What the 2026 Directions are and who they cover
The Directions carry the references RBI/DoS/2026-27/410 and DoS.CO.CSITEG.4/31.01.015/2026-27 and are dated 31 July 2026. The RBI page shows the text as “Updated as on October 01, 2026”. They are organised in eight chapters: the Board’s role, IT governance, IT and information-security risk management, baseline cybersecurity and resilience requirements, the Cyber Security Operations Centre (CSOC), IS audit, and repeal.
- Effective: immediately on issue (para 2).
- Applies to: commercial banks — banking companies other than small finance banks, payments banks and local area banks; corresponding new banks; and the State Bank of India (para 3).
- Foreign banks in branch mode: read “Board” as the controlling or head office, and apply a “comply or explain” approach to Chapters II, III, IV and VII and to listed paragraphs of Chapter V (para 4).
- Repeal: the existing cybersecurity framework and IT governance directions for commercial banks are repealed (para 230).
How this differs from the CERT-In Directions
Banks are covered by both sets of rules. Keep them apart.
- Reporting: RBI para 182 asks banks to report cyber incidents within six hours of detection on DAKSH, the Reserve Bank’s Advanced Supervisory Monitoring System, and also to proactively notify CERT-In. CERT-In’s 2022 Directions separately ask for reporting within 6 hours of noticing an incident.
- Log retention: RBI paras 93–97 require audit trails that support audits, forensics and non-repudiation, plus regular log monitoring, but set no fixed retention period. The bank decides scope, frequency and storage after consulting stakeholders (para 97). CERT-In sets a rolling 180 days within Indian jurisdiction.
- For CERT-In’s 180-day log rule and how to configure it on AWS, see our CERT-In log retention guide.
The clause-to-AWS control map
Each row names the RBI paragraph, summarises what it asks, and lists the AWS service or setting that produces evidence for it. AWS services support the control; the policy, review and sign-off remain the bank’s.
| Criterion | What it asks (summary) | AWS service or setting | Evidence to keep |
|---|---|---|---|
| Para 47 (inventory) | Up-to-date inventory of information assets with business criticality | AWS Config resource inventory; tagging policy; AWS Organizations account list | Config inventory export; tag compliance report |
| Paras 66–67 (secure configuration) | Documented baselines for all device categories; periodic review | AWS Config rules and conformance packs; Security Hub CSPM standards | Conformance pack results; exception register |
| Paras 93–96 (audit logs) | Audit trails on systems touching critical data; enough detail to identify events; regular monitoring | Organisation CloudTrail trail; CloudWatch Logs; VPC Flow Logs; S3 Object Lock on the log archive | Trail and log-source inventory; Object Lock configuration; monitoring runbook |
| Para 97 (log scope and storage) | Bank sets scope, frequency and storage after consulting stakeholders | Retention policy applied to CloudWatch Logs and S3 Lifecycle | Approved log policy; retention settings per log group |
| Paras 99–100 (patching) | Risk-based patch inventory; track patch status for servers, databases and applications | Systems Manager Patch Manager and Inventory | Patch compliance reports |
| Paras 105, 109–111 (privileged access) | Centralised authentication; least privilege; mandatory MFA for privileged users of critical systems; log privileged activity | IAM Identity Center with MFA; service control policies; IAM Access Analyzer; CloudTrail | MFA enforcement evidence; access reviews; privileged session logs |
| Paras 106 and 140 (encryption) | Protect data at rest and in transit; strong, current algorithms | AWS KMS customer managed keys; AWS CloudHSM; TLS policies on load balancers | Key policies; TLS policy settings; encryption coverage report |
| Paras 143, 212–223 (CSOC) | CSOC with log collection, SIEM correlation, alerting and 24x7 monitoring | Logs and findings (GuardDuty, Security Hub CSPM, Inspector) routed via EventBridge or S3 to the bank’s SIEM | Data-source onboarding list; alert rules; shift rota |
| Paras 151, 154 (VA/PT) | VA at least every 6 months and PT at least every 12 months for critical and DMZ systems; approach covers cloud-hosted systems | Amazon Inspector for continuous vulnerability scanning; independent PT under AWS’s penetration-testing policy | VA/PT reports; closure tracker reported quarterly (para 161) |
| Paras 165, 167, 169 (DR and backup) | Half-yearly DR drills for critical systems, a full working day on DR; restore-tested backups | AWS Backup cross-Region copy (Mumbai and Hyderabad); AWS Backup restore testing; AWS Elastic Disaster Recovery for server replication; infrastructure as code for an identical DR site | Drill reports; restore test results; RTO and RPO measurements |
| Para 182 (incident reporting) | Report within 6 hours of detection on DAKSH; notify CERT-In | GuardDuty and Security Hub CSPM findings to EventBridge, SNS and on-call; findings exported to S3 | Detection timestamp; DAKSH submission time; CERT-In notification |
| Para 211 (forensics) | Forensics and DDoS mitigation support on standby | EBS snapshots and isolation runbooks; AWS Shield Advanced with Shield Response Team access (needs a Business or Enterprise Support plan) | Forensic runbook; standby support contract; Shield Advanced subscription and SRT access settings |
| Paras 132–134 (third parties) and Outsourcing Directions, 2025 paras 5, 53(2), 92 | Right to audit, RBI access and data-location rules; cloud governance, SOC log integration and exit strategy | AWS Artifact reports; India Regions; documented exit plan | Cloud provider due-diligence file; contract clauses; exit plan |
Logs, monitoring and the CSOC
Paras 93–96 ask for audit logging on every application or system that can access or affect critical or sensitive information. Logs must carry enough detail, such as date, timestamp, source and destination, to identify each event; audit trails must support audits, forensic evidence and dispute resolution; and logs must be monitored regularly.
Paras 143 and 212–223 ask for a CSOC for continuous surveillance. Its systems must collect logs, correlate them through SIEM tools, monitor dashboards and raise alerts (para 215); it must support root-cause analysis, collection of indicators of compromise and forensics (216, 219); and it needs round-the-clock Level 1 monitoring, with Level 2 and Level 3 expertise (221). Para 223 allows in-house staffing or managed-service arrangements.
On AWS, an organisation CloudTrail trail, CloudWatch Logs, VPC Flow Logs and service logs feed the bank’s SIEM, and GuardDuty and Security Hub CSPM findings go through EventBridge to the SIEM and to on-call. Where the 2025 Outsourcing Directions apply (material cloud outsourcing), they add that logs and events from the cloud provider should be integrated into the bank’s SOC, and that relevant logs should be kept in the cloud for incident handling (para 92(xi)).
Identity, privileged access and encryption
Paras 104–113 ask for access only on a valid business need; privileged users closely supervised and logged; centralised authentication with least privilege and separation of duties; MFA based on risk, and mandatory MFA for privileged users of critical systems and activities (para 110); dormant accounts deactivated; and abnormal logon patterns monitored. Para 106 asks banks to protect data at rest and in transit, and para 140 asks for strong, non-deprecated algorithms, key lengths and protocols.
On AWS, that points to IAM Identity Center with MFA, AWS Organizations service control policies, IAM Access Analyzer, CloudTrail for privileged activity and AWS KMS customer managed keys, with AWS CloudHSM where keys must sit in single-tenant HSMs under the bank’s control. Where the Outsourcing Directions apply, they ask for encryption keys and HSMs to be under the bank’s control in container-based data management (para 92(vii)(b)).
Configuration, patching and VA/PT
Paras 66–67 ask for documented security baselines for every device category, with periodic reviews. Paras 98–103 cover patch and change management: impact assessment, approvals, rollback, and root-cause fixes after incidents.
Paras 149–161 ask for VA and PT on all critical and internet-facing systems: VA at least every six months and PT at least every 12 months for critical systems and DMZ systems with a customer interface (para 151); a documented VA/PT approach that also covers systems hosted in the cloud (para 154); independent, trained testers (para 155); and closure status reported to the IT Strategy Committee and Information Security Committee at least quarterly (para 161).
On AWS, AWS Config rules and conformance packs track baselines, Systems Manager Patch Manager reports patch status, and Amazon Inspector scans EC2 instances, container images in Amazon ECR and Lambda functions for vulnerabilities continuously. AWS lets customers run penetration tests on listed services, such as EC2, RDS, Lambda and API Gateway, without prior approval; DoS testing is prohibited and some simulated events need a request form.
Backup, DR drills and recovery
Paras 163–174 ask for a BCP and DR policy aligned to good practice such as ISO 22301; DR drills at least half-yearly for critical systems (para 165), run from the DR site for at least a full working day (para 167); data backed up and periodically restored to check it can be used (para 169); near-zero RPO for critical systems (para 171); and identical configuration and patch levels at the primary and DR sites (para 173).
On AWS, that means AWS Backup with cross-Region copy between Mumbai (ap-south-1) and Hyderabad (ap-south-2), AWS Backup restore testing for scheduled restore checks, AWS Elastic Disaster Recovery where servers need continuous replication to the DR Region, and infrastructure as code to keep the two sites identical. Hyderabad is an opt-in Region, so it has to be enabled in each account before use. For the Mumbai–Hyderabad design itself, see our AWS Backup and DR guide.
Incident response and DAKSH reporting
Paras 175–186 ask for a cyber incident response and recovery policy covering classification, roles (including outsourced staff), reporting channels, communication, containment, threat-information sharing, post-incident review and testing, with forensic analysis where needed (para 181). Para 182 asks banks to report cyber incidents within six hours of detection on DAKSH and to proactively notify CERT-In.
Paras 192–193 ask for a Board-approved Cyber Crisis Management Plan covering detection, containment, response and recovery, and para 211 asks for network forensics and DDoS mitigation support on standby. On AWS, Shield Standard protection is automatic; Shield Advanced adds access to the AWS Shield Response Team for DDoS events, which needs a Business or Enterprise Support plan. Forensics support is still the bank’s own arrangement.
A checklist for the six-hour clock:
- Before any incident: DAKSH access and named reporters in place; the CERT-In contact route in the runbook; incident classification aligned to the bank’s policy (para 175(1)).
- Detection (T+0): record the detection time, because the RBI clock starts at detection (para 182); open an incident record; name an incident lead.
- First hour: contain by isolating affected instances or accounts (para 180); preserve evidence with snapshots and confirm the log archive is immutable.
- Hours 1–4: pull CloudTrail, VPC Flow Logs, GuardDuty, Security Hub CSPM and Inspector findings for the affected accounts and time window; draft the report.
- Before T+6h: submit on DAKSH; notify CERT-In; log both submission times.
- After filing: find and fix the root cause (paras 102, 181); escalate to the Board and senior management under the communication plan (para 183); feed lessons back into the response plan (paras 178, 187).
Third parties and the cloud provider
Paras 126–135 cover IT and cybersecurity third-party arrangements that fall outside the Managing Risks in Outsourcing Directions, 2025: vendor risk assessment, the bank’s accountability, a right to audit and an RBI right of inspection in contracts (para 132), RBI access to information even if infrastructure is off-premises (para 133), and compliance with rules on infrastructure location and cross-border data movement (para 134).
The Managing Risks in Outsourcing Directions, 2025 (RBI/DOR/2025-26/171, 28 November 2025) apply their IT chapter to a bank’s material outsourcing of IT services, and list cloud computing services and managed security services as IT outsourcing (para 5(1)(v)–(vi)). IT outsourcing is material if a disruption or compromise could significantly affect the bank’s business operations, or if unauthorised access to, loss or theft of customer information could have a material impact on customers (para 53(2)). Existing IT outsourcing agreements had to comply on renewal or by 10 April 2026, whichever came first (para 2).
For banks using cloud, para 92 adds cloud-specific duties: a cloud adoption policy, risk assessment of the cloud provider, controls under the shared-responsibility model, identity and access management with MFA (92(viii)), encryption keys and HSMs under the bank’s control (92(vii)(b)), cloud provider logs integrated into the bank’s SOC (92(xi)), cloud provider resilience and DR drills (92(xv)), and a documented exit strategy (92(xvi)).
Put simply: if a bank’s AWS use is material, the Outsourcing Directions govern it, and the Cybersecurity Directions’ third-party paragraphs (126–135) cover IT and cyber arrangements outside them. The materiality call is the bank’s. AWS publishes compliance reports, such as ISO and SOC reports, through AWS Artifact, and has an India page for financial institutions on its Compliance Center. These help due diligence; they do not make a bank compliant.
Beyond India
The method in this guide — take each clause, name the control that meets it, and decide what evidence proves it — works for any regulator’s rules on AWS. The clause numbers and deadlines here are specific to the RBI.
Where Upvantix fits
Upvantix is an AWS Select Tier Services Partner. We can help a bank’s technology team review its AWS accounts against the clauses in this map: logging and SIEM feeds, privileged access and MFA, configuration baselines, vulnerability scanning, backup and restore testing, and the evidence for each. We are not RBI, are not a CERT-In empanelled auditor, do not act as the bank’s independent VA/PT auditor or IS auditor, and do not give legal or regulatory opinions.
This guide summarises the public text of the RBI Commercial Banks Cybersecurity Directions, 2026 (as shown updated on 1 October 2026), the RBI Commercial Banks Outsourcing Directions, 2025 and AWS documentation, as read on 10 October 2026. It is general information, not legal or regulatory advice. Directions and AWS features change; confirm your obligations with your compliance and legal teams and the current RBI website.
Common questions about the RBI Cybersecurity Directions on AWS
Which banks do the RBI Cybersecurity Directions, 2026 apply to?
Commercial banks: banking companies other than small finance banks, payments banks and local area banks, plus corresponding new banks and the State Bank of India (para 3). Foreign banks in branch mode follow a “comply or explain” approach for listed parts (para 4).
When did the Directions take effect?
Immediately on issue, 31 July 2026 (para 2). The RBI page shows the text as updated on 1 October 2026.
How fast must a bank report a cyber incident to RBI?
Within six hours of detection, on RBI’s DAKSH platform, and the bank must also proactively notify CERT-In (para 182).
Do the RBI Directions set a log-retention period like CERT-In’s 180 days?
No fixed period is set in paragraphs 93–97. Banks must keep audit trails that support audits, forensics and dispute resolution, and decide scope, frequency and storage after consulting stakeholders. CERT-In’s separate 180-day rule still applies.
How often must banks run VA/PT and DR drills?
For critical systems and customer-facing DMZ systems: VA at least every six months and PT at least every 12 months (para 151). DR drills for critical systems at least half-yearly (para 165).
Can a bank run its CSOC through a managed service provider?
Para 223 allows in-house staffing or managed-service arrangements for 24x7 monitoring. The bank remains accountable for outsourced arrangements (para 127). Managed security services are listed as IT outsourcing in the 2025 Outsourcing Directions, so a material managed-SOC arrangement also falls under their IT chapter (paras 5(1)(vi) and 53(2)).
Does Upvantix certify RBI compliance?
No. Upvantix is an AWS Select Tier Services Partner that helps review and configure AWS controls and evidence. Compliance decisions sit with the bank, its auditors and its advisers.