Skip to content
compliancebase
ISO 27001A.8 — Application security requirements

A.8.26

Application security requirements

ISO 27001 · ISO/IEC 27001:2022 · Last verified July 2026

Objective

Identify, specify, and approve information security requirements for applications.

Points of focus

  • Security requirements are captured as part of feature specs or ticket templates, not a separate afterthought process
  • Requirements cover authentication, authorization, input validation, and data handling per classification
  • Requirements for third-party integrations and APIs address authentication and rate/abuse limits
  • A named approver (security engineer, tech lead) signs off on requirements for higher-risk features
  • Requirements are traceable to the resulting acceptance criteria and tests

Implementation notes

Bake a lightweight security-requirements section into the templates engineers already use — the ticket, the RFC, or the design doc — rather than standing up a separate security-requirements process that competes for attention with the normal workflow. For any feature touching customer data, ask three questions up front: who should be able to access this data and under what role or scope, what should be logged when it's accessed or changed, and what happens with data the user doesn't own (cross-tenant isolation is the single most common source of severe SaaS vulnerabilities). For new or modified API endpoints, specify the authentication scheme, expected rate limits, and input validation rules as part of the design rather than leaving them to be inferred from the framework's defaults. Apply the same rigor to internal tools and admin panels as to customer-facing surfaces — these often hold broader data access with weaker scrutiny and are a frequent finding in real incidents. For higher-risk features (new authentication flows, payment handling, bulk export capabilities), require a named reviewer's sign-off on the requirements before implementation begins, and keep that approval visible in the ticket history. Finally, close the loop by making sure acceptance criteria and test cases actually reference the stated security requirements, so 'done' includes 'the security requirement was verified,' not just 'the feature works.'

Audit tip: Pull up the ticket or spec for a recent feature that reads or writes customer data and ask the engineer to point to where the authorization and data-handling requirements were specified before the code was written — not the pull request description written afterward to explain what was built.

Evidence auditors typically request:

  • Feature spec or ticket template with a required 'security requirements' or 'security considerations' section
  • A sampled ticket for a customer-data-facing feature showing authorization rules and data-handling notes
  • API design doc specifying authentication scheme, rate limiting, and input validation for a new endpoint
  • Sign-off or approval comment from a security reviewer on a higher-risk feature's requirements
  • Acceptance criteria or test cases that map back to the stated security requirements

Common gaps

  • The ticket template has a security section but engineers routinely leave it blank or write 'N/A' without anyone checking
  • Security requirements exist for the core product but are skipped for internal tools and admin panels that also touch customer data
  • API endpoints ship without documented rate limiting or input-validation requirements, discovered only after an abuse incident

Cross-Framework Mapping

FrameworkRequirementImplementation note
ISO 27001A.8.26This control
SOC 2CC8.1Related Trust Services Criteria themes — map in your crosswalk; not identical requirements.
GDPRArticle 25Related GDPR articles for personal-data security or processor themes — not a compliance claim.

Primary sources

Frequently Asked Questions

No. Reserve formal review for features touching authentication, authorization, payments, or customer data; routine UI or internal tooling changes can use a lighter checklist or none at all.

A.8.26 is about specifying what the system should do securely (requirements); A.8.27's architecture principles and threat modeling are about how you design the system to meet those requirements and catch what could go wrong.

Tenant isolation is the top priority — every data access path should be scoped to the requesting tenant by construction (e.g. enforced at the query layer), not just by application logic that can be bypassed by a bug.

Yes for anything that handles customer data or authentication — document the expected data flow, what the vendor can access, and any authentication or encryption requirements before integrating.

Framework versions referenced in this page:

  • ISO/IEC 27001ISO/IEC 27001:2022

Last verified: July 2026 · Primary sources linked above