Using AI safely
What not to type into AI, how to check its mistakes, and how to spot fake videos and voice calls.
AI is a powerful tool with its own rules. Remember these five.
1. Don’t type confidential data
What you write into free AI services is stored on the company’s servers and may sometimes be used to improve its models. So don’t enter:
- passport details, card numbers, passwords;
- customer lists, phone numbers;
- internal financial data, contracts.
If needed, replace names and figures: “Client A”, “amount X”. Companies can use business plans where data isn’t used for training.
2. Check important facts
AI can be confidently wrong. Always check numbers, dates, laws, and medical or legal advice against the original source. Tools that cite sources, such as Perplexity, make this easier.
3. Respect copyright
Before using AI-generated images or music commercially, read the service’s terms. Some free plans don’t allow commercial use. Don’t use well-known brands, characters or real people’s likeness without permission.
4. Watch out for deepfakes
AI can now fake anyone’s voice and face. If a “relative” or “boss” calls urgently asking for money:
- hang up and call them back yourself;
- ask something only the two of you would know;
- never read out an SMS code.
5. You are responsible
An email written by AI goes out under your name. Code written by AI runs on your site. Always read it through, and don’t publish what you don’t understand.
In short
AI is a fast assistant, not a judge. Keep secrets out, check what matters, make the decision yourself.
CC BY-SA 4.0