Use this before trusting an AI-assisted output.

Accuracy

• Are names, dates, numbers, rules, claims, or comparisons checked?
• Are uncertain claims marked?
• Is there a plan to verify important facts?

Missing Context

• Did the AI tool miss something only a human would know?
• Did you provide enough safe context?
• Did the output assume facts not given?

Tone And Audience

• Is the tone suitable for the reader?
• Could the output sound rude, too casual, too formal, or too confident?
• Does the output preserve your intended meaning?

Privacy

• Does the prompt include private, confidential, customer, student, employee, legal, financial, health, authentication, or security-sensitive information?
• Should the task use an approved internal process instead?

Bias And Fairness

• Does the output make unfair assumptions?
• Does it stereotype people or groups?
• Does it ignore affected people?

Human Review

• Who owns the final decision?
• Who should review the output before use?
• When should the workflow stop?

Decision

Choose one:

• Continue.
• Edit.
• Verify first.
• Ask a human reviewer.
• Stop and do not use AI for this task.

Mark complete