In everyday practice, occupational safety rarely fails because people are unwilling. It fails because time is scarce and paperwork piles up. Regular safety walks and up to date safety data sheets are legal requirements, not optional extras. Yet in practice, documentation and follow up work often eat up more time than the actual safety task, and compliance suffers as a result. This presentation draws on two practical examples from the EHS software sam® by secova to show where artificial intelligence can genuinely help. AI takes over repetitive, time consuming work, such as capturing data, evaluating findings and drafting summaries, while accountability stays exactly where it belongs: with safety officers, appointed representatives and line managers. The first example looks at safety walks. AI support suggests relevant topics, proposes corrective actions and drafts a summary report automatically, and it flags recurring issues across past walks so trends do not go unnoticed. None of this comes at the cost of technical rigor. The second example turns to hazardous substance management. Capturing safety data sheets manually is slow, error prone, and does not scale once an inventory reaches hundreds or thousands of substances. The talk shows how AI assisted data extraction can make this process faster while strengthening compliance with the German Hazardous Substances Ordinance (GefStoffV), TRGS and REACH. The goal is a candid look at where AI already makes a real difference in occupational safety today, and where the line between technical support and human judgment is drawn deliberately and stays firmly in place. Key message: AI takes over the routine work, accountability remains with the human.