Government of Quebec: How to use artificial intelligence safely
Utiliser l’intelligence artificielle de façon sécuritaire
AI summary
The Government of Quebec released a guide on the safe use of artificial intelligence, flagging risks such as data privacy, content bias, hallucination, legal compliance, copyright and code security. The guide notes that AI systems often collect and retain data, that users may unintentionally transmit personal or confidential information, and that AI-generated code may contain security vulnerabilities; attacks against AI models include data poisoning, data extraction and jailbreak, i.e., bypassing guardrail mechanisms to generate normally restricted content. Facts should be verified before using or sharing AI-generated content.