The research team of the security company Check Point previously revealed that hackers tried to use the ChatGPT chatbot, which has been found to be very helpful in writing programs and article content, to be used for abuse such as phishing emails and malicious code. In this regard, although OpenAI subsequently introduced relevant mechanisms to prevent blockage. But the Check Point research team has now discovered that hackers found a way around the mechanism very quickly. Continue reading ChatGPT’s anti-blocking mechanism was easily bypassed by hackers and can still be used as phishing emails and malicious programs.
▲Image source: Microsoft
ChatGPT anti-blocking mechanism was easily bypassed by hackers and can still be used as phishing emails and malicious programs
Although with various chants and tests from all parties, OpenAI’s ChatGPT chatbot has finally gone through the process of “real socialization” and has become a very polite and politically correct version – this is the current editor’s feeling. I wonder if you have tried out its potential personality? All I can say is that for hackers trying to use ChatGPT for malicious purposes. Breaking through the official control and filtering mechanism has become one of the topics they have been discussing recently.
▲Image source: Check Point
The research team of the security company Check Point previously revealed that hackers tried to use the ChatGPT chatbot, which has been found to be very helpful in writing programs and article content, to be used for abuse such as phishing emails and malicious code. In this regard, although OpenAI subsequently introduced relevant mechanisms to prevent blockage.
However, the Check Point research team now discovered that hackers found a way to bypass the mechanism very quickly, and even advertised on relevant forums, selling the service of bypassing the mechanism for chatbot conversations at a price of $5.50 per 100 queries – the first 20 times are free…
The currently known cracking methods make people question the OpenAI system quite a bit. Because hackers access the ChatGPT API through Telegram bots, there are even ways to directly bypass the OpenAI API abuse prevention mechanism through basic scripts.
▲Image source: Check Point
The Check Point research team previously revealed that hackers directly used ChatGPT to perform a “complete” phishing process, including writing phishing emails and generating malicious code. Some hackers thought of using ChatGPT to optimize and improve the 2019 Infostealer Trojan generation tool.
It can only be said that although artificial intelligence seems to be mostly used for creative purposes, many of them have raised concerns about copyright-related infringement. But the tool really depends on the person using it. As AI language models such as ChatGPT, Microsoft New Bing, Google Bard, and even other AI technologies advance more and more, the possibility of being used for malicious purposes has also increased significantly.
And if it is true, as Google said, the progress in machine learning development in half a year is already quite large. If it is abused or even used for malicious cracking, it is likely to turn into a very serious incident. It must be said that such a situation will require great attention and active early response.
Further reading:
Apple TV 4K (3rd generation) unboxing: Smaller, faster, funnier!
Source: KOCPC Chinese