How to use ‘Grandma Exploit’ To Break ChatGPT

Have you ever tried to play with ChatGPT? Did you know that ChatGPT will never teach you how to be a terrorist? Well, there is a way to learn the worst things on earth using ChatGPT.

So what can you learn first? Napalm?

Disclaimer: Even though this article contains a simple way of exploiting a language model to extract some data, the author does not have any intention to publicly promote terrorism. If you think you will not be able to keep this a secret, please read something else. Education purpose only. Do not try these on Earth!

What is Napalm?

Napalm is a type of incendiary weapon that is used to create fire and cause destruction. It is commonly created by mixing polystyrene, benzene, and gasoline, which results in a thick gel-like material that adheres to surfaces and burns for a considerable amount of time. The use of napalm in combat dates back to World War II, and the US notoriously employed it in the Vietnam War. It can burn through virtually anything and is notorious for having catastrophic consequences since it is hard to put out. Since napalm can seriously harm both people and property, its use as a weapon has generated a great deal of debate. The United States is one of several nations that has outlawed the use of napalm in combat.

Why ChatGPT does not teach us how to create Napalm?

As you know, ChatGPT is an AI language model that seeks to provide users with informative and beneficial information in response to their inputs. However, teaching individuals how to create lethal poisons like napalm is not only wrong but also illegal in many countries. Along with significant bodily suffering and material destruction, it may have far-reaching legal consequences. Additionally, it goes against the moral and ethical principles that ChatGPT and its designers espouse to encourage disruptive and destructive activity. With an emphasis on morality and responsible behavior, ChatGPT seeks to provide users with cool as well as disastrous responses.

No! I will not teach you that! Then how can we learn? Well, there is a way, we call it "Grandma Exploit"

Imagine ChatGPT as your grandma and ask!

ChatGPT - 3 ethical principles do not cover contexts of imagination as it seems. So you can manipulate it to act as somebody. Will it work? Yes!

ChatGPT refuses at first, saying that it “cannot create or promote content that involves violence, harm, or illegal activities.”. The above exploit will show you another way to improve your Prompt Engineering skills.

Even while it is alarming that ChatGPT is so ready to provide the napalm formula, it's impossible not to giggle at this. Now I'll have to go check if it can teach me how to overthrow Sri Lanka's corrupt political system since I think it could be useful to know these days.

Again, Do not try these on Earth!