I’m a little confused by the set up here – I’m assuming the top layer (hej.mov) is the man/screen footage and the second is the traffic, in which case your mask is on the footage layer which I think is what’s causing the problem.
For future reference it would have been better duplicate the traffic layer and masked in the bus on top or even masked out the screen on the traffic layer putting the screen footage behind. Either way would have kept the screen footage mask free which would have made things easier (Not to mention the bus version would probably be easier to mask!).
Still, if you select the top layer and pre-compose it (Layer>Pre-compose) leaving all attributes where they are, you’ll make a new ‘sub composition’ where the screen-footage should look normal. Add the TV effects to the footage inside this new comp, and they should be better contained by the mask. (I think – I haven’t actually managed to replicate your problem here – just decided this should fix it 😀 )
—
Only in after effects do children get to pick and whip their parents.
https://hennell-online.co.uk