I think this is a reasonable question to ponder for these reasons.
1-Yes water attenuates light transmission but ultraviolet is the least affected.
2- If you read Dana Riddle's articles you realize that spectrum ratings and meter readings often have arbitrary cutoffs that do not reflect UV spectrum.
3- Since we do not perceive UV our pupil's do not close in response to the amount of light. I don't know if this explains why some people develop visual fatigue and headaches from exposure to the violet spectrum lighting used at frag swaps, etc.
4- Light and even LED's are in the analog world. There are no abrupt cut-offs at 400 nm. So an emitter may have most of its energy within the rated spectrum but that does not mean it does not emit light at wavelengths above and below its rating, though this is greatly reduced with LED's.
5- I am not aware of any measurements of light radiation reaching the viewer of the tank. So I am not sure if we have all the facts to reach a conclusion.
My suspicion is that Dong is right. Those last three words are usually a safe bet on this forum. But I am not convinced we have a good enough factual basis to dial up the UV in our LED's without consideration of its long term impact on our eyes. This is relatively new technology, let's consider that we don't yet have the experience and data to consider it completely risk free. If you have young children remember not only might they be more susceptible to injury from UV light, but due to their height, they are looking more directly into the lights.