Verification checks AI output or another artifact against evidence, requirements, or trusted signals before it is used. It protects accuracy, privacy, and accountability because AI can produce subtle errors, including nonexistent API endpoints or mixed information. section 6.2 section 7.5
A practical verification routine includes:
- Check important AI responses against reliable evidence instead of accepting plausible wording at face value. section 7.5
- Record the AI model, processing date, and person who verified the result; label processed sections and retain versions. section 6.2
- Use extra review for sensitive or high-impact documents, and remove confidential details from prompts. section 6.2 section 7.5
Verifying a suspicious email
Check the sender domain, SPF, DKIM, and DMARC results, links, urgency, external-sender warnings, and the Received-header route. A mismatched domain such as sharepoint-docshare.com, failed authentication checks, and pressure to click a Review button justify caution. section 7.1.3
Table 6.2 Ethics in AI document handling
|
Ethical concern
|
Best practices
|
|---|---|
|
Data privacy
|
Only use AI tools your company approves for sensitive files. Skip free public AI services for work tasks.
|
|
Transparency
|
Tell your team when you use AI on shared documents. Be open about how AI helped.
|
|
Human oversight
|
Check AI’s work yourself, especially for important documents. AI helps but doesn’t replace your judgment.
|
|
Bias awareness
|
Watch for unfair or slanted results. AI can show bias or make mistakes, just like people do.
|
|
Intellectual property
|
Be careful when creating content with AI. Make sure you’re not accidentally using someone else’s work.
|
|
Accountability
|
Own your work, even when using AI. Don’t blame the AI if something goes wrong.
|
|
Continuous learning
|
Keep up with AI best practices. The field changes fast, so stay current on using AI ethically.
|
Use the checklist to review privacy, transparency, human oversight, bias, intellectual property, accountability, and continued learning. section 6.2
