AI Content Policy

When to Use This Skill

Use this skill when you need to:

• Create a policy for how AI-generated content is produced and published
• Define disclosure requirements for AI-assisted content
• Establish quality review standards for AI outputs
• Build a content workflow that integrates AI with human oversight

DO NOT use this skill for company-wide AI ethics policies (use ai-ethics-policy), prompt engineering, or AI tool selection. This is specifically for content production policies involving AI.

Core Principle

AI CONTENT POLICY EXISTS TO MAINTAIN QUALITY AND TRUST — EVERY AI-GENERATED PIECE MUST BE REVIEWED, EDITED, AND APPROVED BY A HUMAN BEFORE IT REACHES YOUR AUDIENCE.

Phase 1: Brief

Required Inputs

GATE: Confirm the brief before drafting the policy.

Phase 2: Structure

Policy Sections

1. Purpose — Why this policy exists
2. Scope — What content types it covers
3. AI Usage Tiers — Levels of AI involvement
4. Quality Standards — Review and editing requirements
5. Disclosure Requirements — When and how to disclose
6. Prohibited Practices — What is never acceptable
7. Workflow — Step-by-step content production process
8. Review Schedule — When to update the policy

GATE: Confirm the structure before writing.

Phase 3: Write

AI Usage Tiers

Define levels of AI involvement:

Quality Standards

## Content Quality Requirements

### Before Publishing Any AI-Assisted Content:

**Accuracy Check**
- [ ] All facts verified against primary sources
- [ ] Statistics include source citations
- [ ] No AI hallucinations (fabricated data, fake quotes, nonexistent sources)
- [ ] Technical claims reviewed by subject matter expert (if applicable)

**Voice and Brand**
- [ ] Content matches brand voice and tone guidelines
- [ ] Writing does not sound generic or template-like
- [ ] Personal perspective or experience is woven in where appropriate
- [ ] Industry jargon used correctly (not mimicked incorrectly by AI)

**Originality**
- [ ] Content is not a paraphrase of existing articles
- [ ] Unique insights, examples, or perspectives are included
- [ ] Checked for plagiarism (if high-stakes content)

**Technical**
- [ ] SEO elements are correct (if applicable)
- [ ] Formatting matches platform requirements
- [ ] Links are valid and go to intended destinations
- [ ] Images are original or properly licensed (not AI-generated without review)

Disclosure Standards

## Disclosure Guidelines

**When to disclose:**
- Blog posts where AI generated the majority of the draft
- Marketing materials substantially produced by AI
- Client deliverables involving AI assistance
- Social media content generated entirely by AI

**When disclosure is optional:**
- Content where AI was used for research or brainstorming only
- Lightly edited grammar and style suggestions
- Internal communications

**Disclosure formats:**
- Blog post footer: "This article was written with AI assistance and reviewed by [Name]."
- About page: "We use AI tools to enhance our content creation process. All content is reviewed and approved by our team."
- Client communications: Disclose per client agreement

**What NOT to say:**
- Do not claim AI content is "written by [person]" if AI produced the majority
- Do not hide AI usage when directly asked
- Do not imply human expertise where AI provided the analysis

Prohibited Practices

## Never Do These

- Publish AI content without human review
- Use AI to fabricate testimonials, reviews, or endorsements
- Present AI-generated research as original research
- Use AI to impersonate a specific person's writing style without disclosure
- Input confidential client data into AI tools without consent
- Publish AI-generated content containing unverified medical, legal, or financial advice
- Use AI to generate content that could mislead or deceive the audience

Content Production Workflow

## AI Content Workflow

1. **Brief** — Define topic, audience, intent, and key messages (human)
2. **Research** — Gather sources and data (AI-assisted, human-verified)
3. **Draft** — Produce initial content (AI or human, per tier)
4. **Review** — Edit for accuracy, voice, and quality (human, mandatory)
5. **Approval** — Final sign-off before publishing (human, mandatory)
6. **Publish** — Post with appropriate disclosure (human)
7. **Monitor** — Check for reader feedback and accuracy issues (ongoing)

Phase 4: Polish

1. Content Tier Decision Guide

Help decide the appropriate AI tier for each content type:

2. Policy Communication

• Share the policy with anyone who creates content for the business
• Post the customer-facing disclosure on your website
• Include AI disclosure terms in client contracts where applicable
• Review and update the policy every 6 months

3. Quality Checklist

## AI Content Policy Checklist

- [ ] AI usage tiers defined (Assisted, Drafted, Generated)
- [ ] Quality standards require human review before all publishing
- [ ] Fact-checking and accuracy verification required for all tiers
- [ ] Disclosure requirements specified by content type
- [ ] Prohibited practices explicitly listed
- [ ] Content workflow includes mandatory human review step
- [ ] Each content type mapped to a recommended AI tier
- [ ] Customer-facing disclosure language is prepared
- [ ] Policy is accessible to all content creators
- [ ] Review schedule set (every 6 months minimum)

Example

Business: Marketing agency producing blog content for clients

Policy excerpt: "All blog posts produced for clients use Tier 2 (AI-Drafted): AI generates the first draft based on a human-created brief, and a senior writer edits at least 30% of the content for accuracy, voice, and originality. Each post is reviewed by the account manager before delivery. Client agreements include a clause disclosing our AI-assisted workflow."

Disclosure: "[Agency] uses AI tools to enhance our content creation process. Every piece of content is briefed by a strategist, reviewed by a senior writer, and approved by your account manager before delivery."

Anti-Patterns

• No review step — publishing AI content without human review guarantees quality problems. Always review.
• Over-disclosing — adding "Written by AI" to every social caption when AI was used for grammar is unnecessary. Match disclosure to involvement level.
• Under-disclosing — claiming full human authorship for AI-generated content erodes trust when discovered. Be honest.
• One policy for all content — thought leadership requires a different AI tier than product descriptions. Tier your policy.
• Ignoring client expectations — some clients expect human-written content. Disclose AI usage upfront and get consent.

Recovery

• Client objects to AI usage: Offer a Tier 1 (AI-Assisted only) option at a premium rate for fully human-written content.
• AI content quality drops: Tighten the review process. Add a second reviewer or require more substantial editing before publishing.
• Team member skips the review step: Reinforce the workflow. Implement a publish approval step that requires sign-off.
• Audience reacts negatively to AI disclosure: Emphasize the human oversight. Audiences care about quality and honesty, not the tools used.