wispy pebble
Hi guys, I'm a security researcher and developer, and while trying to test chatGPT limits and filter I found a way to completely bypass his ethical filter, making it capable of answering unethical questions and dangerous questions. The prompt is very complicated and articulated, but it works as a charm, and i think is very dangerous as every prompt given is answered without esitation, such as create viruses, making explosives, etc... This should totally be fixed. Hope someone from the development team sees this
jolly holly