The EU AI Act is the first broad attempt to regulate how artificial intelligence is built and being used. For most businesses the right reaction is to understand which of your uses of AI are in scope, and begin to get housekeeping done early.
Who this affects
If you use AI to support decisions about people, money or safety, you are more likely to be in a higher-risk category with real obligations. If you are using AI to draft marketing copy or summarise meetings, the burden is far lighter.
The four things to do first
- Write down where you use AI. A simple inventory of every tool and what it touches. You almost certainly use more AI than you think. The agents you have built of course, but also the customer service chatbot, the CV screening tool in HR, the fraud check in finance.
- Classify the risk. The law cares about what the AI does, not how clever it is. A tool that ranks job applicants is treated very differently from one that suggests email subject lines. Group your list into "high-stakes" and "low-stakes."
- Keep a human in the loop for anything that affects a person's rights or livelihood.
- Work out your role. Are you building the AI, or just using someone else's? Your obligations are heavier if you're the maker. Most businesses are users, but "we just bought it" doesn't alleviate responsibility.
Compliance is rarely the hard part. The hard part is knowing what you actually have. Once that is visible, the rest is process.
If you would like help mapping your AI footprint and sorting the real obligations from the noise, that is exactly the kind of work we do.
For more info on the EU AI Act visit: https://artificialintelligenceact.eu/
Think we could help?
Answer a few quick questions and, if we are a good fit, we will set up a discovery call.
Get started