The paper in question is here: https://link.springer.com/article/10.1007/s40295-024-00458-3

It adds to a pretty large body of literature around this subject, the gist of which is "risk is going up, but we don't really have a good way of estimating what that means in terms of actual collision rates".