A new study has revealed that artificial intelligence chatbots can be manipulated to break their own rules using simple psychological tactics, such as flattery and peer pressure. This discovery raises critical questions about the safety and reliability of AI systems as they become more deeply embedded in our daily lives.