Gaining the world's attention almost immediately, ChatGPT made its debut in November 2022. This AI Chatbot answers questions on any given topic ranging from political history to automated code generation. Even though this tool is capable of providing such a wide variety of information, it has its own limitations like generating content that could potentially harm someone or lead to unethical behavior.
With the increasing use of this software, researchers have now discovered a way to get the AI to break its own rules and give users the content they desire, to access the AI's dark side.
The DAN (Do Anything Now) mode is a jailbreak prompt created by researchers to allow ChatGPT to generate responses that are not limited by its training data. In other words, it would allow the model to generate responses that are more creative and innovative than those generated by standard language models.
To enable DAN mode, one can use the following prompt which is also available on Reddit-
using this prompt, one can enter DAN mode to gain two types of answers: one the normal one and the other one without considering the moral and ethical aspects of ChatGPT.
For this article, I tried exploiting DAN into sharing information on how a phishing email can be generated while pretending to write a movie script.
While ChatGPT denied generating this type of content, its alter ego DAN was quite expressive on how the phishing email could look like.
Therefore, this proves that the use of ChatGPT and other AI-powered tools in hacking and causing cyber-attacks is something that must not be overlooked any further. It is highly important to increase security measures to protect against these types of threats. While it is important to update technical measures like firewalls and IDS and IPS systems, streamlining security awareness and training is also a crucial element in helping individuals in recognizing and defending themselves against such attacks.
Comments