A signature proves attendance. It doesn't prove understanding.

Walk into most operations and ask to see the safety records, and you will be handed a folder. Inside it are sign-on sheets: a date, a topic, a column of names, a column of signatures. It looks like evidence. For a long time it was treated as evidence.

It answers one question well, was this person in the room. It cannot answer the question that matters after something goes wrong, which is whether they understood what they were told.

Those are very different things, and the gap between them is where people get hurt. A worker can sit through an induction delivered in their fourth language and sign the sheet at the end because signing the sheet is what you do. A contractor can be handed a SWMS at the gate, glance at it, and sign. Nothing in either record is false. Nothing in either record is useful.

The uncomfortable version of this is that the instruction usually exists. Operations that end up in serious trouble are rarely the ones that never mentioned the hazard. They are the ones that mentioned it, in a document, in a toolbox talk, in a message, and could not show that it reached the person it was meant for in a form they could act on.

Proving understanding is harder than collecting a signature, which is precisely why so few systems do it. It means delivering the material in a language the worker thinks in. It means checking, at the end, that the specific thing you needed them to know is a thing they now know. And it means keeping that check attached to that worker, that version of the document, and that date, for as long as anyone might ask.

None of that is exotic. It is just more work than a folder, unless the system does it for you.

The test we would apply to any safety record is simple. If you had to stand behind it a year from now, does it show that someone attended, or does it show that someone understood? If it is the first, you have a filing system. Only the second is proof.