Security researchers have discovered a new highly effective jailbreak that can force nearly every major large language model ...