I don't expect the circuit to work well for
three four possible reasons:
The schematic is wrong, can you clarify where the 470k resistor is. The transistor E-B junction is drawn shorted out. An
MPSA29 Darlington has an emitter-base voltage up to 2V so when it's on, the ATTiny would at best see 3V for a logic one. You have internal pullup on or off for that input pin?
You have to software debounce the water level signal sometimes several seconds due to splash and electrical noise. A small capacitor across the transistor's input can prevent RF pickup.
If the water tank not stand-alone, i.e. a sump in the ground, or piping to water from/to the earth, then the tank water has a (current) path to earth-ground. Ther result is a ground potential between your project's 12V power supply and the water tank. It would corrode the electrode on Port 1 as 5VDC flows to earth ground through a complicated path, whenever the electrode is wet, and can cause false triggering.
edit: a last reason for false triggering can be dissimilar metals in the tank, the electrode metal can make a galvanic potential enough to turn on the transistor even though the DC power electrode is dry. Example aluminium, zinc, galvanized metals are all over is -0.75V (chart from structx.com)