It seems that this may be due to the limitations of machine vision. When the body scanners were being deployed, there was a lot of concern that they could see under clothes and that the TSA officials were basically going to be able to see everyone naked as they walked through the scanner. To deal with that, it was decided to use machine vision to create a stylized "Ken Doll" body and superimpose areas of concern, but not allow the actual image to be seen.

Now, with certain hairstyles, I would guess the algorithms can't reliably distinguish hairstyle from concealed items and just alert. I would guess that a human looking at the the imaging would be able to distinguish, but that would involve the TSA being able to see you naked again.