If there were no trend, then there would still be new records established over time, because the longer we keep records for, the more time there is for rare events to occur.

So I think that in order to argue that new records are evidence in favor of a trend, one needs to quantify the likelihood of it not being due to chance somehow.