I don't take it from who is saying what that it causes a lot of cancer. Safe is not a technical term. It's a judgement call. If you could simulate it perfectly, you could decide at what level of glyphosate use more people die from using it than would from side effects of not. Physiology is averaged out by wide sampling in studies, so you have that tool. But people will still disagree on the tradeoff between food production and cancer risk.

A lot of organizations are concerned with one thing: making money, or making food, or preventing cancer. They just come up with the answer that prioritizes what they care about.