Interestingly, Latin ursus and Greek arktos are cognates. Both come from the Proto-Indo-European word for "bear", h₂ŕ̥tḱos.

This word is interesting in that it contains an example of what's called (for various reasons) a "thorn cluster". Certain words in PIE containing the cluster "tk", for whatever reason, underwent metathesis (switching places) in most of the IE daughter languages. This is why the PIE word has a "tk", but the Greek word has a "kt".

This is one of the many reasons for thinking that the Anatolian branch of Indo-European (Hittite, Luvian, etc.) branched off from PIE first - the Hittite word for "bear", ḫar-tág-ga-aš, still shows the PIE order of "t" and "k" (the Hittite double-g was probably something similar to a "k" in this environment, and what appears to be an intervening "a" is a shortcoming of Hittite's cuneiform writing system), meaning that this family of languages branched off before the rest of the family underwent this shared change of "tk" to "kt".

Another fun fact about the "bear" word is that all of Germanic has completely lost it. Instead, in prehistoric times they innovated a formation meaning "the brown one", which is still reflected in Modern English bear.

This is thought to have been due to taboo avoidance. When you're hunting the bear (or maybe when the bear is hunting you), you don't want to actually say the true name of the animal, because that would either scare it away or bring it to you, whichever is worse under the circumstances. So, you instead call it "the brown one" so as not to draw its attention, and so, over time, the true word for "bear" in Germanic was completely lost.

A similar process may have happened with the source of one of the primary Slavic words for "bear", medved <*medu-ed "honey-eater" (the first part cognate with English "mead", and the second with "eat").