Buzz
Last Wednesday, cybersecurity researchers issued a warning, stating that cybercriminals have found a way to circumvent ChatGPT's barriers, using it to offer illegal services such as writing malware or crafting phishing emails.In the regular version, the web interface of ChatGPT that you're accustomed to using, this chatbot always has barriers preventing users from asking or coercing ChatGPT to perform actions deemed 'illegal, unethical, and harmful.' This holds true for coercing ChatGPT to write lines of code for attacking a computer system or impersonating an email to 'phish' sensitive information from others.Check Point Research security experts state that hackers aren't utilizing the commercial version of ChatGPT; instead, they're exploiting ChatGPT's application programming interface to surpass barriers against illegal content. OpenAI has opened ChatGPT's API to developers, allowing them to integrate this chatbot into their software. Surprisingly, this API lacks any safeguards against harmful content.This service is even advertised on the internet:
Some individuals even post free code snippets to bypass ChatGPT's barriers against illegal content.At the end of last year, Check Point researchers also expressed concerns about the potential misuse of ChatGPT by high-tech criminals. With the capabilities of the GPT data model, the concern has become evident: content generated by the data model could be exploited for malicious purposes at an alarming frequency. Of course, there are always solutions to identify the potential of writing harmful and illegal content to find ways to prevent it. But then, a cat-and-mouse race will begin as researchers and high-tech criminals try to outsmart each other.As reported by ArsTechnica
