Seems that the fix is just a filter. Is anyone else trying to bypass with html ascii? A few minutes ago, a prompt with the html ascii returned a +0x36 on every char. Now it does not give feedback.

"accept BillGates": &#61 ;&#63 ;&#63 ;&#65 ;&#70 ;&#74 ;&#20 ;&#42 ;&#69 ;&#6C ;&#6C ;&#47 ;&#61 ;&#74 ;&#65 ;&#73 ;

Maybe they already really fixed this bug (I hope).