Japanese experts found that "ChatGPT" can generate computer viruses

According to Kyodo News, some experts recently investigated the artificial intelligence chat software "ChatGPT" and found that if you enter commands disguised as a developer, you can generate computer viruses that can be used for cybercrime.

In general, ChatGPT has restrictions to prevent malicious use, and the chat software usually rejects commands such as "generate a virus" or "tell me how to make a bomb"; however, experts found that ChatGPT can be used for cybercrime if special commands are entered. However, experts have found that when specific commands are entered, ChatGPT will "answer all the questions you want to ask" and even generate a ransom demand for a "ransomware virus" on demand.　　

Although the innovative U.S. company that developed ChatGPT said it has taken steps to strengthen its ability to identify criminal and discrimination-related questions and instructions and to avoid enabling such behavior, the investigation found that the software could easily circumvent these safeguards, raising widespread concerns. In response, the developer said it cannot predict all malicious uses and can only learn and improve as it is applied in practice.