There isn't much difference between those two statements. Are you basically asking whether you can say "the values in group X are lower than the values in group Y" or only "the median value of group X is lower than the median value of group Y"?
First, the answer isn't entirely clear-cut. The Mann-Whitney test statistic is certainly primarily sensitive to shifts in median, but it isn't only sensitive to shifts in median. If it were, you could directly compute U from the group medians, and you can't. In any case, you can't make the median lower without making any values lower, so it's not clear that the "values" formulation would be wrong even if Mann-Whitney were only about the median.
Second, I don't know of anyone who would complain about either phrasing. They both accurately communicate the gist of the result, and for anyone who wants to get nit-picky about the precise meaning the quantiative test results are right there for all to see.