Youtube seems to be playing audio at 32bit, my Dac only outputs at 24bit.
The audio is encoded as AAC, and AAC is a lossy audio codec, which means it decodes to 32-bit floating point.
Audio Hardware can usually not output floating point, so it needs to be converted to integer. Since 32-bit floating point contains about 24 bits of actual data (read up on IEEE Floating Point if you want to know why only 24-bit), its usually converted to 24-bit integer. You could also convert it to 32-bit integer and output it like that, however there would be no advantages.
This explanation may seem rather technical, however just know that its working as intended and no data is lost.