2.4 — When Human Review or Additional Verification Is Required

CCAO-F Domain 2 · Task 2.4

What You Need to Know

Escalation passes a Claude output to a qualified person to review before use. Triage decides who reviews, never whether anyone does.

The Triggers

• External audience (customers, public, press, funders)
• Legal, financial or contractual consequence
• Decisions about people (hiring, appraisal, allocation)
• Safety or health implications
• Irreversibility once sent

One trigger is enough. Confidence and polish are not on the list.

Qualified, Not Senior

Anthropic Usage Policy High-risk AI Recommendations require a qualified person (training or experience in that field) with authority to change the output, before it is delivered as advice or used to make a decision. The person concerned must also be told AI was used. Seniority and availability are neither necessary nor sufficient.

Both Failure Directions

• Under-escalation — sending externally-bound or consequential content because the draft reads authoritatively
• Over-escalation — committee for a rota note; every routine draft through a professional reviewer. Buries reviewers and erodes the channel high-stakes outputs need

Key Concept: Triage asks two questions in order. What does this output carry? If anything, who is qualified in that subject to catch the errors that matter?

Exam Traps

• Choosing the reviewer by seniority or availability
• Waving output through because it is only internal
• Escalating everything to be safe