Most major large language models (LLMs) can quickly tell when they are being given a personality test and will tweak their responses to provide more socially desirable results—a finding with implications ...