If it's still a concern, a low-dropout current limit circuit can be added to limit the current to 1 A. However, a series switch & current sense is going to be a bit tricky on 3 V, a simple fuse or PTC device (like used in cellphone battery) would probably be better.
I was going to suggest a sense resistor in series with the 0V rail and another transistor plus zener diodes to cut-off the the bases when the voltage across the sense exceeds Vbe. Then I realised that when it's short circuited the transistors would still have to dissipated 3V × Ilimit which would exceed the maximum rating of anything in a TO-92 package.
I think 1A is too high anyway, PTC resistors take awhile to act and will allow several times the rated current to pass before cutting out; a typical small transistor e.g. PN2907A has a maximum current rating of 800mA and will probably be destroyed before the polyfuse blows.
I'd recommend using a PTC rated to twice the motor current, even then this might be too much.
giftiger_wunsch,
Does your MCU have a comparator?
I have an idea that might work, you could configure it so it oscillates when the current limit is exceeded, this will prevent the transistors overheating as the duty cycle could be set low.
Look at my previous circuit.
Notice how I've implemented protection from both inputs being high by adding a zener so it shorts Tr2's base to 0V when Tr1 is activated?
Current limiting can be implemented in a similar manner.
The comparator is configured as a Schmitt trigger oscillator with R10 and C1. Normally the oscillator is inactive, when the voltage across Rs exceeds the lower threshold of the Schmitt trigger, Tr7 is turned on which connects the bases of Tr1 and Tr2 turning Tr3 to Tr6 off. With all the transistors off, C1 discharges until its voltage drops below the upper threshold of Schmitt trigger, Tr7 turns on and the cycle repeats.
This current limiter adds extra complexity but it might be worth it.