We know that $\int_0^\infty \cos x \; dx$ is undefined. This seems like a very simple question, but I just want to confirm it is okay to do so:
Can I say that $$\bigg|\int_0^\infty \cos x \; dx \;\bigg |\leq 1 \,\textrm{?}$$
Since it is equal to $$\bigg[\sin x\bigg]^\infty_0 = \lim_{x \rightarrow \infty} \sin x$$
I have genuinely not seen this done before, it seems like we can say it for sure. Since the limit does not exist but it is not "unbounded"?
You are almost correct. We have $$ \int_{0}^{M}\cos x\,dx = \sin M \tag{1}$$ but $\lim_{M\to +\infty}\sin M$ does not exist. In english terminology every non-convergent sequence is usually called "divergent", and this case brilliantly explains why I prefer the italian/french terminology, where we have convergent, divergent and non-convergent sequences: not every divergent sequence (in the english terminology) is actually unbounded, so it is kind of awkward to call them all "divergent".
There also is a way to extend the concept of convergence and to assign to such integral an actual value. For every $\lambda>0$ the integral $\int_{0}^{+\infty}\cos(x)\,e^{-\lambda x}\,dx$ does exist, and it equals $\frac{\lambda}{\lambda^2+1}$. So: $$\text{ER}\int_{0}^{+\infty}\cos(x)\,dx = \lim_{\lambda\to 0^+}\int_{0}^{+\infty}\cos(x)e^{-\lambda x}\,dx = 0 \tag{2}$$ where $\text{ER}$ stands for exponential regularization.