It didn't take much research to find a casement (side/single hung) window opener/actuator. They come in 24V form too.
Given they "claim" up to 400N force, although in cheap AliExpress form they probably only produce half that. It's still 20Kg of force. Even (I think) when you take the angles and torque/moments into account.
A cat tail or paw in the opening or the "jam" and this motor sounds like it could cause pain and even damage.
I found a very old thread elsewhere where someone suggested using a 12V car window motor. The ideal being that they shouldn't be able to provide enough force without stalling out to prevent fingers getting nipped in windows etc. Others had suggested truck wiper motors.... these do not have any concern for fingers and I have been hit by a wiper blade in motion and ... it did make an attempt to nearly slash my face open while physically pushing me out of it's way.
I was considering a complicated current sense shunt and a latch circuit to cut the motor power for a second maybe if it's current hits a value where the motor is going to deliver too much force for a finger in the window jam.
It occurred to me an even simpler limiter would just by a power resistor. If the window opener draws a sensible amount of current, then a sensible amount of voltage will drop across the resistor. If the motor stalls and the current rises more and more will drop across the resistor.
Then it occured to me that the motor and it's windings are ... electrical engineering magic things beyond my understanding.
I'm going to guess, something like: The stalled current spike is caused by the coil reading saturation (the rotar is not moving the magnetic field is not being driven, behaving like low ohm resistor. Thus a series resistor of a fixed value would struggle to limit the current successfully.
Back to the current "trip" circuit. Possibly with, possibly without an automatic recovery.
For the opener to function as a "let the cat out" device it will have to have other automation. Some of it will be in software, some may even be on a server. So there is the capacity for the "current overload" status to "signal" to the automation and the "automatic recovery" could be triggered either by a "power down" or a "reversal". Such that the hardware "lock out" remains in force until there is signs of awareness by intervention.
For the actual automation, that's far more about my cat's particular behaviour than it is electronics. I have a range of sensors available for this, from basic smart buttons under a sheet of plastic that triggers when cat stands on it, to ultra-sonic prox sensors, motion sensors etc. I expect a combination and a bit of fiddling. "IR Beam sensors" are also an option, particularly for determining when the cat is standing IN the window opening.... as they have a tendency to do while deciding.
Other things "nice to have" would be a "false start" when closing. Make the window move suddenly, but only for a tiny distance. This should be enough to spook the cat out the way. Then the window will close 2 seconds later. Relying on the beam sensors to stop it if the cat gets in it and the current limiter as back up such that if the cat is STILL in the gap somehow it only gets a gentle squeeze for a few milliseconds. The window, if this is signaled to the automation, could then reopen, pause for 1 minute and then repeat the attempt to close.
How to tell if it's the cat or a cat burglar? This is worth considering. I'm still working with this.