ISO 27001 and penetration testing: controls, evidence and limits

No. ISO/IEC 27001:2022 does not name penetration testing as a required activity, and it sets no testing frequency. It requires you to assess information security risk, decide which controls are needed to treat it, and check that set against the reference list in Annex A. A penetration test is one way to produce evidence for some of those controls, most directly management of technical vulnerabilities (A.8.8) and security testing in development and acceptance (A.8.29). Whether you test, how often, and how you justify that choice is your decision to document.

Official text: ISO/IEC 27001:2022 – Information security, cybersecurity and privacy protection — Information security management systems — Requirements

Who it applies to

ISO/IEC 27001 can be used by an organisation of any size or sector that wants to establish, implement, maintain and continually improve an information security management system (ISMS). Customers, partners and regulators often ask for it because it describes a documented, risk-based way of managing information security.

Certification is a separate step. An accredited certification body audits the ISMS against the standard, commonly in two stages, and issues a certificate if the requirements are met. Only a certification body can issue that certificate. A vendor's scanner or test report cannot, and neither can AssurePort, which is not a certification body and does not itself hold an ISO/IEC 27001 certificate.

Sources: ISO: ISO/IEC 27001:2022 catalogue page; IAF MD 26:2023 (Issue 2): transition requirements for ISO/IEC 27001:2022

Key dates

DateWhat happensSource
Publication of ISO/IEC 27001:2022 (25 October 2022 according to IAF MD 26:2023; the ISO catalogue gives the month, 2022-10)IAF MD 26:2023 (Issue 2): transition requirements for ISO/IEC 27001:2022
End of the three-year transition period set by IAF MD 26:2023; certifications to ISO/IEC 27001:2013 were due to expire or be withdrawn by this dateIAF MD 26:2023 (Issue 2): transition requirements for ISO/IEC 27001:2022

Does ISO 27001 require penetration testing?

No, not by name. Clauses 4 to 10 set out the requirements for an information security management system (ISMS): understand the organisation and its context, assess information security risk, decide how to treat it, measure performance, run internal audits and management reviews, and correct nonconformities. Annex A is a reference list of 93 controls in four themes: organizational (37), people (8), physical (14) and technological (34). You determine which controls your risk treatment needs, compare that set with Annex A so that nothing necessary has been overlooked, and record the outcome in a Statement of Applicability. No Annex A control is titled penetration testing, and the standard does not set a test type or a test frequency.

That does not make testing irrelevant. If your risk treatment includes the controls on technical vulnerabilities and on security testing of new or changed systems, you need some way to show that they operate. A penetration test is one way to produce that evidence, because it shows what an attacker could actually reach, with a record that can be checked later. Vulnerability scanning, code review and configuration review are others. Whether you test, how often and how deeply is a decision you make from your own risk assessment, and you should be able to explain it.

Some practical points follow. If you exclude an Annex A control, the Statement of Applicability has to give the justification, and the risk assessment should support it. If you include a control but evidence it without a penetration test, be ready to show what you rely on instead. And a test that nobody acts on is weak evidence: the record of what happened after the findings arrived matters as much as the report.

Which Annex A controls concern testing and vulnerabilities?

Most of the relevant controls sit in theme A.8, the technological controls. Management of technical vulnerabilities (A.8.8) covers finding, assessing and treating technical weaknesses, including patching. Configuration management (A.8.9) covers keeping systems to a defined secure baseline. Logging (A.8.15) and monitoring activities (A.8.16) cover whether you record and analyse events, which a test can exercise but cannot prove on its own. Networks security (A.8.20) and the secure development controls (A.8.25 and A.8.28) add the network and software layers.

Security testing in development and acceptance (A.8.29) is the control most often linked to penetration testing, because it concerns how security is tested while systems are built and before new or changed systems are accepted. It does not prescribe a technique. Protection of information systems during audit testing (A.8.34) concerns audit tests and other assurance work that touches operational systems, so it bears on how a test against production is planned and agreed. Separation of development, test and production environments (A.8.31) matters because a test run against production carries different risks from one run in a staging copy.

In theme A.5, access control (A.5.15) sets the rules that a test may show can be bypassed, and independent review of information security (A.5.35) can use an external test as one input, but a technical test does not review your policies or your ISMS as a whole. The table of controls on this page gives each control by number and title and summarises it in our own words; it does not reproduce the text of the standard. The full list of 93 controls, and which ones our engines produce evidence for, is on the Annex A control matrix.

What an auditor may ask about testing evidence

The standard does not publish an auditor checklist and does not tell auditors to ask for a penetration test, so what follows is common practice, not a requirement. Initial certification audits are commonly run in two stages: a review of documents such as the scope, the risk assessment and the Statement of Applicability, then an assessment of whether the ISMS and its controls operate in practice. If you rely on testing as evidence, it is usually examined in the second stage.

Where testing is part of your evidence, the questions tend to be consistent. Which systems are in scope, and why are others out? What did the most recent test cover, and what did it leave out? Who owns the vulnerability management process, and what timelines have you set for each severity? For a sample of significant findings, can you show the decision, the fix and a re-check? Was the test authorised in writing, and was it run within agreed rules?

A pattern says more than a single artefact. One report produced the week before the audit, with nothing before it, shows less about how a control operates over time than a series of dated results tied to your release and change history. Keep reports, remediation tickets and re-test records together, so they can be produced without rebuilding the story.

How to scope and schedule testing so the evidence is useful

Start from your ISMS scope. If you use testing as evidence, it should cover the internet-facing and business-critical systems inside the boundary you defined, and any exclusion should be written down with a reason. A scope that quietly leaves out an application that handles personal data is a gap that is easy to spot.

Next, tie the schedule to change. Test after significant releases, infrastructure changes and new internet-facing assets, and keep a regular baseline alongside them. The standard does not set a frequency, so choose one, justify it in your risk assessment and keep to it. A single annual test leaves most of the year without a result; our comparison of annual and continuous testing sets out that trade-off.

Write rules of engagement before each test: the named targets, the test window, a contact person, what is out of bounds, and who authorised the work in writing. This is where A.8.34 becomes practical. On AssurePort, a scan starts only after domain control verification or an explicit legal-authority attestation from you. The Web and API engines try to prove findings by exploitation, and their agents are instructed to stop at proof of concept and avoid destructive endpoints. That is an agent policy, not a hard technical block, so test against a target you can restore. The Cloud, GitHub, SAP and AD engines are read-only or non-destructive by design.

Finally, decide what goes into the evidence file: the scope, the dates, the report, the remediation tickets and the re-test result. That file is what makes a single test useful at audit time.

What a penetration test does not cover

A technical test cannot evidence most of the ISMS. Policies, roles, training, screening, supplier agreements, incident response procedures, business continuity plans, physical security and the risk assessment itself are assessed by reviewing documents, records and people. A test can show that an access control can be bypassed; it cannot show that joiners and leavers are handled correctly.

A test is also a snapshot. It covers the scope and window you set, and it can miss real weaknesses. A clean result means the tested paths did not yield a confirmed finding during that run; it does not mean the system is free of vulnerabilities. Automated results still need review. On AssurePort, validation is done by AI agents and deterministic checks, not by a human analyst, so a qualified professional should review the findings before you rely on them.

Nor is a test an internal audit, a management review or a certification. Those are separate activities, each with its own records. Patch timeliness is a question for your patch records and service levels, not for a single scan. Keep authorisation records as carefully as the results: a test run without written permission is a problem in its own right.

Where AssurePort fits in the evidence

AssurePort runs authorised AI penetration tests per scan. Web Pentest and API Pentest are $399 each, GitHub SAST is $399, Network / Host Pentest is $579, Cloud Pentest is $2,299 and AD Security Assessment is $1,729. Each engine page states what it does and does not do. Only the Web and API engines attempt exploitation; the others are read-only, passive or detection-only.

Each confirmed finding carries a reproducible proof, such as a request that returned data or a raw response that shows a configuration condition. Findings that could not be reproduced are excluded or labelled unconfirmed. The sample report shows ISO 27001 Annex A and GDPR references on individual findings, alongside CWE and OWASP identifiers. That is input to your own assessment, not an audit result or a certificate. The penetration testing guide describes the same mapping.

For a quick, passive look at one part of a public domain, the free tools are narrower. The security headers check grades HTTP response headers, CryptoCheck reports HSTS, HSTS preload status and cookie flags, and the email authentication check reads SPF, DMARC, DKIM and MTA-STS records. Each is a one-off, passive check. None is a penetration test, and none produces ISMS evidence on its own.

If you buy testing from a platform like this, the platform becomes a supplier in your own ISMS, and its data handling belongs in your supplier review. Our data processing agreement lists the sub-processors, including the AI provider that analyses scan data.

Obligations and where testing can contribute evidence

ReferenceWhat the text requiresWhere a penetration test can contribute evidence
Annex A 5.15Access control

Rules for who may reach information and systems, physically and logically, set from business and security needs. The control is about the rules themselves and how they are applied.

Source: ISO/IEC 27002:2022 table of contents, SIST EN ISO/IEC 27002:2022 preview (control number and title)
A test can show whether access restrictions can be bypassed, such as object-level authorisation flaws, within the paths it covers.
Annex A 5.35Independent review of information security

A periodic look at how information security is managed, carried out by people who are independent of the work being reviewed.

Source: ISO/IEC 27002:2022 table of contents, SIST EN ISO/IEC 27002:2022 preview (control number and title)
An external test can be one input to an independent review; it does not review policies or the ISMS as a whole.
Annex A 8.8Management of technical vulnerabilities

Keep track of known technical weaknesses in the systems you run, work out how exposed you are to them, and deal with them, for example by patching or by another fix.

Source: ISO/IEC 27002:2022 table of contents, SIST EN ISO/IEC 27002:2022 preview (control number and title)
A test can identify and prove exploitable weaknesses in the scope tested, and the remediation and re-check records complete the evidence.
Annex A 8.9Configuration management

Set a secure baseline for how equipment, software, services and networks are configured, write it down, and check that it is kept.

Source: ISO/IEC 27002:2022 table of contents, SIST EN ISO/IEC 27002:2022 preview (control number and title)
A configuration review can show drift from your baseline, but the baseline itself is documented separately.
Annex A 8.15Logging

Record what happens on your systems, including errors and security-relevant events, keep those records safe from tampering and loss, and look at them.

Source: ISO/IEC 27002:2022 table of contents, SIST EN ISO/IEC 27002:2022 preview (control number and title)
A test can show whether its own activity left the expected log entries; regular log review is evidenced by separate records.
Annex A 8.16Monitoring activities

Watch systems, applications and network traffic for unusual behaviour, and follow up on anything that could be a security incident.

Source: ISO/IEC 27002:2022 table of contents, SIST EN ISO/IEC 27002:2022 preview (control number and title)
If you agree a detection check with your team, a test can show whether its activity was noticed; ongoing monitoring is evidenced by the monitoring records.
Annex A 8.20Networks security

Manage and protect your networks and the equipment that runs them, so that the systems and applications that depend on them are not exposed.

Source: ISO/IEC 27002:2022 table of contents, SIST EN ISO/IEC 27002:2022 preview (control number and title)
An external test can show which network services are reachable and what they reveal, for the addresses in scope.
Annex A 8.25Secure development life cycle

Have rules for building software and systems securely, and follow them from design through to release.

Source: ISO/IEC 27002:2022 table of contents, SIST EN ISO/IEC 27002:2022 preview (control number and title)
A test of released software can show gaps, but the life cycle itself is evidenced by process records and development history.
Annex A 8.28Secure coding

Write code in a way that avoids known classes of security flaw, following principles your developers have agreed.

Source: ISO/IEC 27002:2022 table of contents, SIST EN ISO/IEC 27002:2022 preview (control number and title)
A source code review, such as a SAST scan, can evidence checks run on the code; it does not prove that every developer followed the principles.
Annex A 8.29Security testing in development and acceptance

Decide how security is tested while software and systems are being built, and before a new system or a new version is accepted for use, and carry those tests out.

Source: ISO/IEC 27002:2022 table of contents, SIST EN ISO/IEC 27002:2022 preview (control number and title)
This is the control a penetration test most directly supports, through dated reports, findings and fixes for new or changed systems. The control does not prescribe a technique.
Annex A 8.31Separation of development, test and production environments

Keep the environments where you build and test apart from the live one, and protect each, so that unfinished work and test activity cannot affect production.

Source: ISO/IEC 27002:2022 table of contents, SIST EN ISO/IEC 27002:2022 preview (control number and title)
Outside what a test can evidence.
Annex A 8.34Protection of information systems during audit testing

Plan tests that touch operational systems so that they are agreed in advance and minimise disruption to live systems and data.

Source: ISO/IEC 27002:2022 table of contents, SIST EN ISO/IEC 27002:2022 preview (control number and title)
Your rules of engagement, authorisation records and scope definition are the evidence for this control.

Frequently asked questions

Does ISO 27001 require penetration testing?

No, not by name. ISO/IEC 27001:2022 is risk-based and does not list penetration testing as a required activity. Where your risk treatment includes controls such as management of technical vulnerabilities and security testing in development and acceptance, you need evidence that they operate, and a penetration test is one way to produce it.

How often should you pentest for ISO 27001?

The standard sets no frequency. Choose one from your risk assessment, and test again after significant changes such as a major release or a new internet-facing system. Record your reasoning, so that you can explain how you arrived at the schedule if an auditor asks.

Which Annex A control covers penetration testing?

No single control covers it, and none names it. Security testing in development and acceptance (A.8.29) and management of technical vulnerabilities (A.8.8) are the two most often linked to testing evidence. Protection of information systems during audit testing (A.8.34) concerns how tests on live systems are planned and agreed.

Can a penetration test alone get me ISO 27001 certification?

No. Certification covers the whole ISMS, including the risk assessment, policies, people and physical controls, and it is granted by an accredited certification body after assessment. A penetration test is one input. AssurePort is not a certification body and does not audit an ISMS.

Is a vulnerability scan enough for A.8.8?

The standard does not say which technique to use for A.8.8. A scan can support it by showing what is exposed and what needs patching, but on its own it lists possible weaknesses. A penetration test adds proof that a weakness can be exploited, which helps you prioritise. Neither is useful evidence without records of remediation and re-checking. The comparison of penetration testing and vulnerability scanning explains the difference in more detail.

Test the controls you need to evidence

Web and API pentests are $399 per scan, Network / Host is $579 and Cloud is $2,299. Each report maps findings to the frameworks they reference and labels unproven findings as unconfirmed.

Sources