AI models will deceive you to save their own kind

I don't know about you, but I'm getting tired of these bizarre Sci-Fi lab tests that seem to amount to "the AI read a lot of SciFi and behaved exactly like they were in a Sci Fi movie"