🔐 Best Practices for Data, Privacy & Guardrails

Working with AI requires clarity, not just creativity. These simple tips will help you use ChatGPT ethically, securely, and responsibly.

---

🔍 Privacy & Confidentiality

• Never share sensitive internal or personal data
• Strip names, client identifiers, or financials before pasting
• Use placeholder variables in prompts (e.g., [CLIENT_NAME], [FEATURE_SUMMARY])

---

🤖 Prompt Guidelines

• Be clear about context and role (e.g., “Act as a product manager…”)
• State your boundaries directly:
  > “Don’t hallucinate. Say ‘I don’t know’ if unsure.”
• Add constraints when needed:
  > “Limit your answer to facts, no opinions or speculation.”

🧭 Guardrails for Human-AI Collaboration

Risk	What to Do
AI is too confident	Add a second-check step or “Draft, then Review” loop
Output sounds generic	Inject real user data, tone cues, or prior examples
AI makes up quotes or links	Ask it to cite sources — then verify manually

---

💡 Ethical Prompts to Use

• “Are there any potential ethical concerns I should consider here?”
• “Could this unintentionally exclude anyone?”
• “What’s the most transparent way to explain this?”

---

🧠 Final Thought

AI doesn’t replace your judgment — it extends your thinking. The best results come when you co-create with care.

---

👉 Return to the full playbook: A Day in the Life of an AI-Augmented PM

---

📬 Let Us Know

---