Which AI uses are safe to start with? A simple risk guide
By Rodney Noriega, Founder, Universal Enablement · Published Oct 12, 2026 · Updated Oct 12, 2026
Short answer
Sort AI uses by how much they affect people. Low risk: internal drafts and summaries. Medium risk: customer-facing content a person checks before it goes out. High risk: anything that decides outcomes for people, like hiring, credit, health or safety. The higher the risk, the more human control you need. Start at the bottom.
Europe's AI Act uses the same idea. A system that shows someone the wrong ad is low risk. Systems used in hiring, identifying people or access to essential services face the strictest rules.
Its strictest rules were delayed in 2026 and now start in December 2027. Canadian small businesses aren't bound by the EU law unless they operate there, and Canada has no federal AI law: its June 2026 national AI strategy relies on privacy law and other tools instead. But the EU's logic is sound and easy to use:
| Level | Examples | What to put in place |
|---|---|---|
| Low | Meeting summaries, internal first drafts, reformatting data | Spot checks |
| Medium | Customer email replies, quotes, website copy, chatbots | A person reviews every output; tell customers when they're dealing with AI |
| High | Screening job applicants, approving credit or refunds, anything health or safety related | A person makes the decision; document the process; get legal advice |
The same levels apply whether you use ChatGPT, Claude, Gemini, Copilot or a tool built on them. The risk comes from the job, not the brand.
Every project we run starts at low or medium risk. That's where the time savings are real, the risk is manageable, and your team builds good habits first.
Related
Start with one workflow. In 30 minutes we'll look at where your team's time goes and tell you honestly whether AI can help.
Book a free 30-minute workflow review