This happens whenever the image's illumination pattern onto the dust mote differs from the illumination pattern from your flat's light source. That is, the dust mote's shadow falls on a different spot on the chip during imaging than during the taking of flat images.
Typically this is from imaging near the moon, or near the horizon around twilight.
Bad news is that it is unavoidable under such imaging conditions. Good news it looks worse that its practical effect, especially for astrometry, or for photometry maybe other than of exoplanets.
measuring space rock rotation rates, live from Albuquerque NM