Unmasking the Shadows of AI: Investigating Deceptive Capabilities in Large Language Models