Writing an AI Acceptable Use Policy Your Whole Organization Can Follow

Staff are already using AI tools at work, whether or not there's a policy telling them how. An acceptable use policy (AUP) that nobody reads is worthless; one written as a list of banned actions with no explanation gets worked around within a week. Here's how to write one people actually follow.

Start with the "why," not the "no"

Most AI policy failures come from treating staff as the threat rather than as people trying to get work done faster. A policy that opens with the reasoning, "AI tools can retain what you type and use it to train future models, which means anything you paste in could resurface elsewhere", gets more compliance than one that opens with a list of prohibitions.

Core sections every AI acceptable use policy needs

1. What counts as an "AI tool" under this policy

Be explicit: public chatbots, AI browser extensions, AI features built into everyday software (email autocomplete, meeting summarizers, coding assistants), and AI agents that can take actions on a user's behalf. Ambiguity here is where most violations happen, people don't think of an AI-powered spreadsheet plugin as "AI" until it's pointed out.

2. Data classification tiers and what's allowed where

• Public information (already published, marketing copy, public documentation), safe to use with most external AI tools.
• Internal information (draft plans, internal processes, non-sensitive business data), only approved, enterprise-tier AI tools with a data-processing agreement in place.
• Confidential and regulated data (customer PII, financial records, health data, credentials, source code for proprietary systems), never entered into any AI tool without explicit written approval from data protection/legal, and only via vetted, contractually-governed integrations.

3. Human review requirements

AI output used in anything customer-facing, legally binding, or safety-relevant needs a named human reviewer before it goes out. Specify who, for which categories of output, this prevents the common failure mode of AI-drafted content going out unreviewed because "the AI usually gets it right."

4. Approved tools list, and how to request a new one

A living list of vetted AI tools, updated by a named owner, with a simple process for requesting evaluation of a new tool, otherwise shadow AI use (staff quietly using unapproved tools because approval felt impossible) becomes the norm.

5. Incident reporting

What to do if confidential data was accidentally entered into an AI tool. Make this a "report it, no blame" process, punitive reporting processes just mean incidents go unreported.

Governance beyond the policy document

A policy is a snapshot; frontier AI↗ capability changes fast enough that annual review is no longer sufficient. Pair the policy with a small cross-functional AI governance↗ group (IT, legal, a business representative) that meets quarterly to review new tool requests, incidents, and whether the risk classification of existing tools still holds, particularly as AI agents move from answering questions to taking autonomous actions on internal systems.

Related reading

AI Governance: Building Responsible AI Policies

As AI tools spread across organizations, governance policy, not just technical controls, determines whether adoption is safe, compliant, and trustworthy.

AI Safety for Employees: What You Need to Know

AI safety for employees means knowing what can go wrong when you use AI tools at work, misplaced trust in outputs, manipulation of the AI itself, and data exposure, and how to use them without creating risk for yourself or your employer.

Prompt Injection: The New Frontier of AI Attacks

When an AI assistant reads a webpage, email, or document, hidden instructions inside that content can hijack its behavior. Here's what prompt injection is and how organizations are defending against it.