No, this behavior is actually normal. The scope input divider does not go to ground but to approx. 2.8V (once it is adjusted), because center is at VCC/2=2.5V. That way the scope can live with just a single supply rail (5V) and does not need a negative supply (would have made the design more complicated and more expensive). If you short the input then the bottom end of the input divider sits at ~2.8V, the other (top) end sits at 0V, and the output to the op-amp works out to 2.5V due to the resistor values (900k and 100k) used, which the software recognizes as zero. But if you leave the output floating the op-amp input will go to wherever the bottom of the divider sits, i.e. ~2.8V, which is 0.3V higher; since the divider is 1:10 the scope (which does not "know" there is nothing attached) now says the input apparently sits at 10x(2.8-2.5)= approx.3V which is exactly what you see.