Unfortunately, this wasn't just the PDF representation - it was for the disordered regions analysis as a whole (all residues were marked as disordered in the output, even when an inspection of the values indicated those that weren't).
For those interested, it was because we changed from using DISOPRED version 2 to version 3 early last year (2022), and the output was not parsed 100% correctly - this is fixed now.
As a general point, if you find that a feature in Phyre2 is broken, please do let us know, and don't assume that someone else has reported it - this one took over 18 months before it was reported (although we try to test everything, automated testing can only do so much)! E-mailing me will often lead to a quicker response than posting here.