One common belief is also that the computer that reads the signals from the brain and moves the limb doesn't receive orders from the cyberarm's node (which provides at most diagnostic information). Thus you have no way to spoof the neural signal unless you hook directly up to the limb. Though honestly, you ought to get a wireless-less cyberarm anyways because why not? Considering how an uber AI made everyone into puppet slaves about 15 years ago, people may still be worried about that.
You don't have to get the arm to send signals back to that computer though. You just need to get into that computer and send signals to the arm.
With a wireless arm, you aren't under any threat at all unless you decide to connect it to a network that is open to wireless. Look at it this way. Take an old late 80s computer hooked up. No connection to anything other than the monitor and keyboard. Nothing can hack into it from the outside because it isn't connected to anything. You sure as hell can't hack in wirelessly. Now, connect that same computer up to a network and take away it's keyboard were it can only receive commands from another computer (we'll call that 1980s computer the cyberarm and the new computer the translator that is interpreting brain impulses). If the other computer isn't connected to anything else, it still is unhackable. Now, hook that other computer to a network full of computers, one of which has a wireless connection. Now, people can hack in through the wireless connection and have fun with
anything on the network. It isn't the easiest thing in the world, but it can be done.
As for the fear, looking at what people have done through open wi-fi connections today and most people still don't bother to even attempt basic matr...errr...internet security measures like setting up a password for their network, I don't think even that is going to change your average person's sheer laziness.
Generally I go with no wireless command of cyberlimbs mainly due to all the questions wireless command opens up. How does one command a cyberlimb? Can my friend command rig it? Can I command rig my own arm? What if it isn't my arm and just an arm duct taped on? Does the duct tape command arm cost me essence? Why do cyborgs hit essence 0 if all they're doing is just losing body parts (which doesn't cost essence)? My soul doesn't normally care about jumping into drones. Can I just be a human stump and move around with my drone body parts? Can I control myself by jumping into my skillwires since they can control my actions? SR doesn't really discuss these sort of things and I don't want to figure it out whole-cloth.
Well in all fairness, in past editions losing limbs could cause you to lose essence on it's own (deadly damage could even if it didn't result in limb loss). I think the hardest part of controlling cyberware is that being DNI, your command is going to get overridden pretty damn fast. So you spoof a command to the arm, the user thinks no thanks, and its over. Some things would work fairly easily (move finger to hold trigger down), others wouldn't be effective (make a fist) because the user can just relax with a thought. I don't think you could command rig a cyberlimb, they aren't really drones. I do think that you could spoof commands to them all day long. Or edit data being relayed by them and such. The key is that, to be effective, they have to be things that either A.) are done in a very short time span (so the user doesn't just nullify it with a thought) or B.) things that the user won't automatically notice (edited data feeds).
I don't know how RAW it actually is, but I know in at least one Mission, the head of a brothel is said to "control" the actions of a joytoy through their skillwires, and their used to be cyberware allowing the rigger to jump into a human and ride along with complete sensory (although it didn't allow direct control). It all just got simplified to the extreme in SR4/A.
Regardless, I don't think the 'hacking cyberware' thing works, and my mind won't be changed. You all can assume it does all you want, but I'm going to continue as I have and keep ruling it as impossible to do, and if a hacker tries it in a game I run, I'll just give them the diagnostic reports and let them know that's all they can get out of it. Their time would be much better served hacking a nearby Americar and causing it to run over the opposition.
You're free to houserule what you want. We never claimed houserules don't exist. A discussion of how the game
actually works really isn't going to benefit from input that only focuses on your houserules though. Likewise, if someone wants to
actually know what something is, putting forth your houserules as an answer only serves to confuse/mislead the new player. That's just cruel.