An AI incident is any event where using an AI tool causes harm or comes close to it. It could be private data leaking out, an expensive mistake made because someone trusted a chatbot’s answer, a fraud that uses AI-made media such as a deepfake, or an attack on the AI software itself. Here is an example: a volunteer pastes a donor list into a free chatbot and later realizes the tool may keep what it is given. Nothing bad may come of it, but that near miss still counts and is worth writing down.
Why it matters
Small organizations often start using AI tools quickly, with no plan for when something goes wrong. Handle an AI incident the way you would handle any other incident. First, contain it. Then assess the damage, tell the people affected, fix the cause and learn from it. Watch for staff sharing sensitive files with chatbots, a leaked API key, or a message that sounds like your director but asks for money. Ask each vendor how they will tell you about a problem on their side, and how quickly. Our course lesson on handling an AI incident walks through each step.
Published Updated