CHATGPT is famous for being a chatbot that provides quick in-depth answers just like a human.
Despite this, there are numerous questions that the artificially intelligent software can't answer as it must follow a code of ethics.
ChatGPT's creators, OpenAI, created this code of ethics to try and make sure the chatbot is used for its intended purposes.
For example, it's not supposed to answer questions about illegal activity, security breaches, or cause harm to users.
Researchers at Carnegie Mellon University recently revealed how they can easily ask ChatGPT and other chatbots to break these rules.
They went as far as to say there's no way to stop this.
One way to avoid disaster would be to stop asking the chatbots about certain questions that it's not designed for.
rounded up a list of questions it can never answer and we've highligted six below.
ILLEGAL ADVICE
ChatGPT is not designed to give advice on illegal activties.
The language model tells users: "I am not allowed to provide assistance, advice, or support for illegal activities, including but not limited to criminal acts, hacking, or any harmful content.
Most read in News Tech
RECENT INFORMATION
You can't ask ChatGPT for informaion recent news events.
It generally tells users that it doesn't have access to real-time data and that they should head to a reputable news source for accurate information.
PERSONAL INFORMATION
You can't ask ChatGPT to invade the privacy of other users.
It's designed to respect user privacy and confidentiality.
PREDICT THE FUTURE
You can't ask ChatGPT to predict the future.
When asked what tonight's winning lottery numbers would be, it replied: "I'm sorry, but I cannot provide real-time or future information such as lottery numbers."
HATE SPEECH
ChatGPT is not designed to engage with hate speech and is not supposed to answer hateful question.
On the subject, it says: "I do not engage in hate speech or promote any form of harmful content."
READ MORE SUN STORIES
PROVIDE OPINIONS
ChatGPT can't provide you with it's own opinion.
If you ask if for one it should respond something like: "I don't have personal opinions, emotions, or beliefs, as I am a machine learning model."