Responsible AI Use in Business Workflows

AI must not act alone when the consequences of a wrong decision are significant and not easily reversed. This includes: publishing content to a live website, sending communications to clients or external parties, updating financial records, making changes to access permissions, and any action that affects data subject to regulatory obligations. For these actions, the AI proposes and a human decides — without exception.

AI drift occurs when an AI system's outputs begin to deviate from what you'd expect based on your guidelines and context. Signs include: outputs that use slightly different terminology than your brand, recommendations that seem more aggressive or conservative than your usual approach, content that references outdated information, and suggestions that contradict recent decisions your team has made. If outputs feel "off", they probably are — flag them rather than approving them.

If an AI output looks wrong: don't approve it. Reject it with a clear reason — this creates a record and helps the system learn. If you're unsure whether it's wrong or just unfamiliar, ask a colleague before approving. If you see a pattern of wrong outputs on a particular task type, report it to your team lead so it can be investigated systematically rather than just rejected case by case.

Escalate when: an AI output is potentially harmful and you're not sure it was caught by the approval process, when you notice a systematic pattern of incorrect outputs, or when you've been asked to approve something you don't understand well enough to evaluate. Escalation is not a sign of incompetence — it's the correct response when the stakes exceed your confidence. Use the standard incident reporting path for your team.