The problem with "is it safe" questions on the open internet is that there is a bias against answering "yes". Give an ominous warning and you sound smart; give a false assurance and you get egg on your head because someone somewhere will run into trouble.
I have never seen any pushback against collaboration with Russian academics in pure science, and I would consider such pushback closeminded and ill-advised. This does not mean that it does not exist, since academics can be closeminded and ill-advised, but then again you might not want to rely on the opinions of such people in your career anyway.
I am much more loathe to collaborate with Russian institutions, particularly when it comes to ones with official ties (e.g., math contests) or odious administrators (at this point, basically any Russian university). The "lending credibility" effect can be significant here, and even if it is not, by strengthening these institutions, you are reducing the push-factor for Russian students and researchers to emigrate, which is not an effect I would want to have.
In applied project with realistic military applications, I would not collaborate with anyone physically in Russia period. How far these "military applications" reach is a tricky question (networking? coding theory? video compression?), and some amount of eyeballing will be necessary at the boundaries.