4 templates
Safety prompts ready to copy
Templates to spot scams, review suspicious messages and protect your accounts. AI helps you see warning signs, but it doesn’t replace your bank or official channels: if in doubt, contact the organisation yourself through its official website or phone number.
Before you paste it: quick checklist
- Never paste passwords, verification codes or card numbers.
- Copy only the text of the suspicious message, without your own details.
- Ask for a list of warning signs and what to do next.
- If you have a real doubt, call the organisation on its official number.
Possible scam detector
I’ve received this message: [message without personal data]. Point out any typical scam signs you see and what I should check through official channels. Don’t decide for me.
Search with official sources
[Question], in [country], with information from [year]. Use official sources and cite each claim. If sources disagree, tell me. Don’t give medical, legal or financial advice.
Check an account’s security
I use [service or app]. Give me a checklist to review my account’s security and privacy: two-step verification, open sessions and devices, connected apps, recovery email and phone, and what is public. Tell me where each setting usually is and remind me to confirm it in the service’s official help, because menus change.
What to do after a possible scam
I think I’ve fallen for a scam: [what happened, without personal data or account numbers]. Put the first steps in order of urgency (contact the bank through its official channel, change passwords, check accounts, report it) and tell me what evidence to keep. Mention official helplines only if you are sure they exist. Don’t ask me for data.
How to adapt it to your assistant
The template works the same in ChatGPT, Gemini, Copilot or Claude because it only uses plain language. If your assistant supports fixed instructions (custom instructions, Gems or projects), you can store your preferred tone and format there and paste only the variable part. On a free plan, split long requests into two or three shorter messages.