That's actually a problem with all kinds of components. It's why a lot of SMT parts ship in sealed bags with desiccant packs and moisture indicator cards, with instructions for baking the parts if the moisture's risen above a certain level. LEDs may be especially sensitive to moisture ingress due to the encapsulants used.
I was thinking about that as well, but that would require the LED to fail short rather than open circuit. It's common for semiconductors to fail short due to electrical damage, but I'm skeptical that it would be a common failure mode due to moisture getting into the LEDs themselves. That said, I'm not that familiar with the chemistry involved when it comes to the materials used in the LEDs. It also depends on the driver topology.I was thinking the same thing about how would a failed LED cause 4 others near it to light up. That didn't make sense to me, but then I thought of how a pinball machine uses diodes to solve issues with column-row driven switch and coil matrixes. I guess since a LED is one way only it causes some kind of issue when driving the neighbors?