A typical poorly reasoned law student argument.
Method of control doesn't really provide a defence - you'd be hard pushed to prove that the design of the controls made you do something. And if you could prove it, 'brain control' would be no different to a poorly designed control panel.
I suspect that the method of interface doesn't actually matter - brain->finger->weapon isn't significantly different to brain-> weapon.
In both cases you could end up reacting too quickly, and cause a bad outcome.
And insufficient buffer between thought & action has always been a problem in all areas of human activity.
In any case I suspect anything designed with this interface concept would be built to require an explicit trigger (i.e. a deliberate thought rather than subconscious) which negates the defence.
As for holding engineers liable - if you build a tool, you may well find yourself liable if it does something unintended but if it works as intended ultimately liability lies with the user.
Any object can be used for bad purposes - hammer, knife, car, biro(!) - and you don't hold the person who designed it liable when this happens.
Even the majority of 'military' objects can be used for defensive as well as offensive roles - and it gets even more complicated when you consider 'lesser of two evils' type arguments. It's a bit more difficult to argue when it comes to things like nuclear armed ICBMs - though even these can be said to have a 'good' purpose if used purely for deterrent.
So it's the end user who holds the can, not the engineer. Unless of course the engineer made a cockup!