The most believable claim I've seen so far is that DNA might be very slightly conductive. Just barely enough to slightly increase replication errors.

I discount it without further evidence, but sounds plausible.