Many organisations publish an information security policy that tries to be a rulebook for every employee and a control framework for the security team at the same time, and end up with a document staff never read and auditors cannot test. An acceptable use policy (AUP) and an information security policy (ISP) answer different questions. The AUP tells each person what they may and must not do with the organisation's information, accounts and devices, and is enforced through acknowledgement and the disciplinary process. The ISP states what the organisation commits to and which controls its systems and processes must have, and is enforced through management, audit and exceptions. This guide is a clause allocation guide for the core workforce documents: which requirements belong in each, who signs what, how personal devices and AI tools fit, and how both map to ISO/IEC 27001:2022 Annex A controls A.5.1 and A.5.10. It is for CISOs, GRC leads, HR partners and policy owners. The AI-specific version of the same split is in the AI Acceptable Use Policy vs AI Security Policy guide.
Key takeaways
- The ISP binds the organisation, the AUP binds the individual. The ISP is a management commitment about systems and processes; the AUP is a set of personal rules each user accepts before getting access.
- Enforcement mechanism decides placement. A clause you would enforce through HR belongs in the AUP; a clause you would enforce through design review, configuration or audit belongs in the ISP or a standard beneath it.
- The AUP must be acknowledged, the ISP must be approved. NIST SP 800-53 PL-4 expects a documented acknowledgement of rules of behaviour before access; ISO/IEC 27001 expects top management to approve the information security policy.
- Duplication is the commonest defect. When the ISP repeats AUP rules, the two drift apart and employees can point to the weaker wording in a disciplinary hearing.
- Volatile detail lives below both. Approved software lists, password lengths and approved AI tools change often and belong in standards or registers that the policies reference.
Two documents, two jobs
The distinction is old and well established in control frameworks. NIST SP 800-53 Rev 5 separates PL-1 Policy and Procedures (the organisation's security policy, its scope, roles and management commitment) from PL-4 Rules of Behavior (the rules that describe individual users' responsibilities and expected behaviour, acknowledged before access is granted and re-acknowledged when they change). PS-6 Access Agreements adds signed agreements, such as nondisclosure and acceptable use agreements, for individuals who need access. NIST CSF 2.0 GV.PO-01 asks for a cybersecurity policy that is established, communicated and enforced, and GV.PO-02 for one that is reviewed and updated.
ISO/IEC 27001:2022 Annex A makes the same split. A.5.1 Policies for information security covers the information security policy and topic-specific policies, defined, approved by management, published, communicated to relevant personnel and reviewed. A.5.10 Acceptable use of information and other associated assets requires rules for the acceptable use of information and assets, and procedures for handling them, to be identified, documented and implemented. A.6.2 Terms and conditions of employment and A.6.4 Disciplinary process are where the AUP gets its contractual and enforcement teeth.
- Information security policy: written for management, control owners, system owners and auditors. It sets objectives, scope, roles, the risk method, the policy hierarchy and the commitment to meet legal and contractual requirements. It is short at the top and is implemented through topic-specific policies, standards and procedures. See the Policy vs Standard vs Procedure vs Guideline guide for how those layers divide.
- Acceptable use policy: written for every employee, contractor, temporary worker and third party with access. Plain language, readable in ten to fifteen minutes, specific about what is allowed, what is prohibited and what to report. It is a condition of access and is referenced from employment contracts and supplier agreements.
Side-by-side comparison
| Attribute | Acceptable use policy | Information security policy |
|---|---|---|
| Who it binds | Individuals with access: staff, contractors, third parties | The organisation, its managers and control owners |
| Question answered | What may I do, what must I never do, what must I report? | What do we commit to, who owns it, and which controls must exist? |
| Voice | "You must", "you must not" | "The organisation will", "system owners shall" |
| Approval | Policy owner with HR and legal sign-off; often workforce consultation | Top management or the board |
| Sign-off by readers | Individual acknowledgement on joining and on material change | No individual signature; communicated and made available |
| Enforcement | Acknowledgement trail, monitoring, disciplinary or contractual remedies | Management review, internal audit, risk acceptance and exceptions |
| Length | Short; a few pages | Short top-level policy plus topic-specific policies and standards |
| Change cadence | Annual review; faster for new tool classes | Annual review; standards change as technology changes |
| Main framework anchors | ISO 27001 A.5.10, A.6.2, A.6.4; SP 800-53 PL-4, PS-6 | ISO 27001 clause 5.2 and A.5.1; SP 800-53 PL-1; CSF GV.PO-01, GV.PO-02 |
| Evidence an auditor asks for | Acknowledgement records, training completion, disciplinary cases | Approval minutes, review dates, communication records, control mapping |
What each document defends against
The AUP targets risk created by individual behaviour: credential sharing, unapproved software and cloud services, careless handling of classified data, misuse of email and messaging, and failure to report incidents. Its threat model includes negligent and malicious insiders, and it supports investigations by establishing in advance what the person was told. See the Insider Threat Programme Guide.
The ISP targets organisational failure: no one owning a risk, controls never specified, inconsistent standards between business units, and no route for exceptions. Its threat model is systemic. A perfectly behaved workforce still leaks data if the ISP never required encryption at rest or logging; a perfectly specified control set is still bypassed if staff were never told not to forward data to personal accounts. The two documents are complements, not alternatives.
Clause allocation: which requirement goes where
The test is simple: who has to act, and how would you prove a breach? If an individual acts and HR would enforce it, it is an AUP clause. If a system owner or function acts and audit would test it, it is an ISP or standard clause. Many topics need a clause in both, written for different readers.
| Topic | AUP clause (individual) | ISP or standard clause (organisation) |
|---|---|---|
| Passwords and MFA | Do not share credentials; do not reuse work passwords; approve only MFA prompts you started | Authentication standard: MFA required for all remote and privileged access; password parameters |
| Email and messaging | No auto-forwarding to personal accounts; report suspected phishing | Email security standard: filtering, external forwarding blocked by default, retention |
| Software and cloud services | Install or sign up only for approved tools; ask before connecting third-party apps to work accounts | Software and SaaS approval process; application control; OAuth consent settings |
| Data handling | Handle data according to its classification label; no work data in personal storage | Classification scheme, labelling, DLP and encryption requirements |
| Devices | Lock screens; report loss within a set time; keep devices updated; follow BYOD conditions | Endpoint baseline, encryption, MDM or MAM enrolment rules, remote wipe scope |
| Personal use | Limited personal use permitted under stated conditions | Not needed |
| Monitoring | Notice of what is monitored and why | Logging and monitoring standard; DPIA; access to monitoring data |
| AI tools | Use only approved AI tools; data classes that must not be entered; human review of outputs | AI security policy and standards for acquired and built AI systems |
| Privileged access | Use admin accounts only for admin tasks; no browsing or email from privileged accounts | Privileged access standard: PAM, session recording, approval |
| Incident reporting | Report suspected incidents immediately via the stated channel; no penalty for honest mistakes | Incident management policy: triage, roles, notification duties |
| Exceptions | Ask for an exception before deviating; do not self-approve | Exception and risk acceptance process with owners and expiry |
Ready-to-adapt wording is in the Acceptable Use Policy Template and the Information Security Policy Template. This page explains the boundary between them rather than repeating them.
Audience and enforceability
An AUP is only as useful as its enforceability. Three conditions matter in practice.
- The person was told: keep a record of which version each person acknowledged and when. NIST SP 800-53 PL-4 expects acknowledgement before access and again when rules are revised.
- The rule is clear: "must not" and "must" wording with concrete examples. "Users should take care with data" cannot support a disciplinary outcome; "you must not upload Confidential data to any service not on the approved list" can.
- The process is fair: in the UK, the Acas Code of Practice on disciplinary and grievance procedures sets the minimum an employer should follow, and an employment tribunal takes the employer's procedure into account. Acas notes that what counts as gross misconduct can depend on the organisation's own rules, so state in the AUP (or the disciplinary policy it references) which breaches may be treated as gross misconduct.
The ISP has a different kind of enforceability. It is enforced against managers and functions through objectives, audit findings and risk acceptance. A system owner who cannot meet a standard does not face discipline for that alone; they raise an exception with a named risk owner and an expiry date. See the Security Exceptions and Risk Acceptance Guide.
Proportionality supports enforceability. The NCSC's 10 Steps guidance on engagement and training says policies and processes should be fit for purpose and proportionate, that people with knowledge of local working environments should be included in policy making, and that workarounds are a sign the controls, not the people, need fixing. An AUP clause nobody can follow in real work will be ignored and will weaken the clauses that matter.
Sign-off and attestation
| Event | AUP | ISP |
|---|---|---|
| Approval | Policy owner (often CISO) with HR and legal; workforce consultation where required | Board or executive committee; minuted |
| New joiner | Acknowledge before first account activation, ideally as a gated step in onboarding | Made available; awareness training explains it |
| Contractor or supplier staff | Acknowledge directly, or supplier contract obliges the supplier to bind its staff | Supplier security requirements reference it |
| Material change | Re-acknowledgement by all in scope | Re-approval and communication |
| Annual cycle | Re-attestation, often bundled with awareness training | Management review and update |
| Leaver | Reminder of continuing confidentiality duties | Not applicable |
Store acknowledgements in a system that records the person, the document version, the date and the method, and that can produce a report of people with access who have not acknowledged the current version. That report is the single most useful AUP control metric.
How personal devices and AI tools fit
Personal devices: BYOD is where the AUP and ISP most often collide. The NCSC's BYOD guidance notes that the organisation owns the corporate data on the device but the user owns the device, that balancing control with users' privacy expectations can be difficult, and that a poorly designed scheme may push staff to shadow IT. The individual conditions (enrolment, screen lock, reporting loss, consent to selective wipe of work data) go in the AUP or a BYOD agreement it references. The technical requirements (which management approach is used, what data is reachable from unmanaged devices, wipe scope) go in the ISP's endpoint or mobile standard. See the BYOD Policy Template and the BYOD Security Programme guide.
AI tools: add a short AI section to the core AUP (approved tools, prohibited data, human review of outputs) and point to a fuller AI AUP if you have one, rather than leaving AI out of the document staff actually acknowledge. The controls on AI systems themselves belong in an AI security policy under the ISP. The AI Acceptable Use Policy Template and the Shadow AI Policy Template provide wording.
Monitoring: an AUP that says "all activity may be monitored" is a notice, not a lawful basis. In the UK the ICO's guidance on monitoring workers sets out employers' obligations under the UK GDPR and the Data Protection Act 2018; the ICO states (checked November 2026) that the guidance is under review following the Data (Use and Access) Act. Keep the plain-language notice in the AUP and the monitoring design, DPIA and access rules in the ISP's logging and monitoring standard. See the Logging and Monitoring Policy Template.
Worked scenario: one incident, two documents
A project manager sets up automatic forwarding of work email to a personal webmail account to read it on holiday. Months later the personal account is compromised and client data is exposed.
- If only the ISP covers it: the ISP says "email shall be protected against unauthorised disclosure". The manager never read it and acknowledged nothing specific. HR has little to act on, and the root cause review finds that the email platform allowed external auto-forwarding because no standard required it to be blocked.
- If only the AUP covers it: the AUP says "you must not forward work email to personal accounts", and the manager acknowledged it. HR can act, but the same incident recurs with someone else, because nothing required the platform to block the behaviour.
- With both: the AUP clause and acknowledgement establish the individual's duty; the ISP's email standard requires external auto-forwarding to be blocked by default, with exceptions approved and logged. The forwarding rule never takes effect, and the attempt is visible in logs.
The lesson generalises: every AUP prohibition should be checked for a technical control in the ISP that makes the behaviour hard, and every ISP control that depends on user behaviour should have a matching AUP clause.
When to merge and when to separate
Small organisations (under roughly 50 staff, no certification) can combine them into one document with two clearly separated parts: a one-page management statement and a user rules section that each person signs. Keep the signature tied to the rules section only. Organisations pursuing ISO/IEC 27001 certification or with regulated data should separate them: the ISP is a controlled management document reviewed by top management, and the AUP is a workforce document with its own acknowledgement cycle, HR ownership of enforcement and, in some jurisdictions, consultation duties. Large and multinational organisations often keep one global ISP and localise the AUP for employment law and works council requirements. See the ISO 27001 Certification Guide and the Security Policy Framework Guide.
Detection and measurement
- Acknowledgement coverage: percentage of active accounts belonging to people who have acknowledged the current AUP version. Target 100%, with access blocked for new joiners until acknowledged.
- Time to acknowledge after change: median days for the population to re-acknowledge after a material revision.
- Clause-to-control coverage: percentage of AUP prohibitions with a matching technical or procedural control in the ISP or a standard.
- Duplication count: number of rules stated in both documents with different wording, found in the annual review.
- Exception volume and age: open ISP exceptions and the share past expiry.
- AUP-related incidents: incidents with a policy breach as a contributing factor, by clause, to show which rules need a technical control or clearer wording.
Common failure patterns
- AUP duplicating the ISP: user rules copied into the ISP and then edited separately, so two versions of the same rule disagree.
- Unenforceable wording: "should", "where possible" and "take reasonable care" in clauses that need to support disciplinary action.
- No acknowledgement trail: the AUP is on the intranet but nobody can show who accepted which version, so it cannot be relied on.
- Contractors left out: third-party staff get accounts through a supplier route that skips the acknowledgement step.
- Lists in the policy: approved software, approved AI tools or password lengths written into approved policy text, so every change needs board approval and the lists go stale.
- Prohibition without control: AUP bans a behaviour the platform still allows by default, leaving discipline as the only control.
Framework mapping
| Framework | Reference | Document |
|---|---|---|
| ISO/IEC 27001:2022 | Clause 5.2 Policy; Annex A A.5.1 Policies for information security | ISP and topic-specific policies |
| ISO/IEC 27001:2022 | Annex A A.5.10 Acceptable use of information and other associated assets | AUP |
| ISO/IEC 27001:2022 | Annex A A.6.2 Terms and conditions of employment; A.6.4 Disciplinary process | AUP enforcement |
| NIST SP 800-53 Rev 5 | PL-1 Policy and Procedures | ISP |
| NIST SP 800-53 Rev 5 | PL-4 Rules of Behavior; PS-6 Access Agreements | AUP and acknowledgement |
| NIST SP 800-53 Rev 5 | AT-2 Literacy Training and Awareness | Communicating both |
| NIST CSF 2.0 | GV.PO-01, GV.PO-02 | ISP establishment and review |
Practitioner checklist
- Inventory every rule in your current policies and tag each as individual (AUP) or organisational (ISP or standard).
- Remove user rules from the ISP and replace them with a reference to the AUP.
- Rewrite AUP clauses as "must" and "must not" with examples, and state which breaches may be gross misconduct.
- Move approved lists and technical values into standards or registers referenced by both documents.
- Make AUP acknowledgement a gated onboarding step for employees and contractors, with version tracking.
- For each AUP prohibition, confirm a technical or procedural control exists in the ISP or a standard.
- Add BYOD, AI tool and monitoring notices to the AUP, with the matching design rules in ISP standards.
- Have the ISP approved by top management and record review dates; have HR and legal sign off the AUP.
- Report acknowledgement coverage and exception age to the security steering group quarterly.
Standards and references
- NIST: SP 800-53 Rev 5 Security and Privacy Controls for Information Systems and Organizations
- NIST: Cybersecurity Framework 2.0
- NCSC: 10 Steps to Cyber Security, Engagement and Training
- NCSC: Device Security Guidance, Bring Your Own Device
- ICO: Employment Practices and Data Protection, Monitoring Workers
- Acas: Code of Practice on Disciplinary and Grievance Procedures
- Acas: Disciplinary Procedure Step by Step
Related NHI Mgmt Group resources: Acceptable Use Policy Template, Information Security Policy Template, AI Acceptable Use Policy vs AI Security Policy, Policy vs Standard vs Procedure vs Guideline, Security Policy Framework Guide.