Some digging in src/mactex.lisp yields the following which seems to work. First remove the TEX and TEX-RBP (right-binding power -- has to do with precedence) properties. Then nuke the function which special-cases trig-like functions for TEX-MEXPT (handles exponents).
(%i1) :lisp (mapcar #'(lambda (x) (setf (get x 'tex) nil) (setf (get x 'tex-rbp) nil)) *tex-mexpt-trig-like-fns*)
(%i1) :lisp (defun maybe-tex-mexpt-trig-like (&rest a) (declare (ignore a)))
Let's see what Maxima thinks is the list of trig-like functions.
(%i1) :lisp *tex-mexpt-trig-like-fns*
(%SIN %COS %TAN %SINH %COSH %TANH %ASIN %ACOS %ATAN %ASINH %ACOSH %ATANH)
OK, let's go through them and see what TeX output we get.
(%i1) for f in [sin, cos, tan, sinh, cosh, tanh, asin, acos, atan, asinh, acosh, atanh]
do tex(f(x)^2);
$$\sin \left(x\right)^2$$
$$\cos \left(x\right)^2$$
$$\tan \left(x\right)^2$$
$$\sinh \left(x\right)^2$$
$$\cosh \left(x\right)^2$$
$$\tanh \left(x\right)^2$$
$$\arcsin \left(x\right)^2$$
$$\arccos \left(x\right)^2$$
$$\arctan \left(x\right)^2$$
$${\rm asinh}\; \left(x\right)^2$$
$${\rm acosh}\; \left(x\right)^2$$
$${\rm atanh}\; \left(x\right)^2$$
(%o1) done
I guess the \; output for asinh etc. is a bit of a wart. We can fix that too if you need it.
The tex output code is an agglomeration of heuristics which work most of the time. As you can see, it's not always obvious how to get a different output.