Templates
Free AI Safety Policy Template
By Get AI Safety Certified Team, Get AI Safety Certified · September 2026
Key Takeaways
- Safety = protect people from the model (accidental harm: bias, hallucinations, leaking PII, over-reliance) — guardrails. Security = protect the model from attackers (adversarial: prompt injection, breaches, exfiltration). Intent is the key distinction.
- Treat every prompt as a message leaving the building.
- A deployer owns what staff paste and what the model tells customers.
What never goes in a ChatGPT prompt
An AI use policy template is useless if it does not name the data that never goes in a prompt: source code, customer files, credentials, health or student records, payment data, and anything under NDA. If you would not print it on a postcard, do not paste it into a public model.
In 2023, reporting described Samsung semiconductor staff pasting confidential source code into ChatGPT. The company then restricted generative AI tools. The lesson for every deployer is egress: a prompt can leave the building even when nobody meant to leak.
In Moffatt v. Air Canada (2024), a chatbot invented a bereavement-fare policy. The Civil Resolution Tribunal treated the airline as responsible for that statement. The company still owns what the model tells a customer.
Module 0 includes this template with the free classroom videos. Write and quiz stay after login. Unlock M1–M5 for $199 when you want the rest of the path. Email is optional — the PDF is not gated.
Email is optional. The PDF downloads either way — no gate in front of the template.