Practical guidance written by our team, last updated Oct 15, 2026. It's general advice, not legal advice.
1. Check the facts
AI can state wrong things confidently (a “hallucination”). Check anything important (health, money, law, news) against a trusted source, and ask the AI for its sources, then open them.
2. Protect privacy
Don't paste passwords, ID numbers, medical or financial details, or other people's private information into chatbots. Check whether your chats are used for training and turn it off if you prefer. For sensitive work, consider running AI on your own computer, so nothing leaves your device.
3. Disclose AI use
Tell people when AI wrote or heavily shaped something you share, especially at work, at school, or when someone expects your own words. Don't use AI voices or images to pretend to be a real person.
4. Respect copyright and creators
Don't present AI output as your original work where that matters, don't ask AI to copy someone's paid content, and check a tool's terms before using its images, voices or music commercially. Only clone a voice with clear consent.
5. Watch for bias
AI learns from human-made data and can repeat stereotypes. Be careful using it for decisions about people (hiring, lending, grading), and have a human make the final call.
6. Keep AI agents on a short leash
Agents can send, buy and delete. Start with read-only access, require approval for anything that sends money or messages, and watch for prompt injection. Use our free AI agent safety checklist (PDF) and see Is it safe to let an AI agent use my email?
New to the terms? See the AI glossary. Want to help others do this well? Join the community and take the pledge.