Man accused of using generative AI to create virus in Japan 1st

TOKYO — A 25-year-old man has been served a fresh arrest warrant for allegedly creating a computer virus using generative artificial intelligence (AI), the Metropolitan Police Department (MPD)’s cybercrime control division announced on May 28, in what is believed to be the first such case in Japan.

Ryuki Hayashi, an unemployed resident of the Kanagawa Prefecture city of Kawasaki, was served the warrant on suspicion of making electronic or magnetic records containing unauthorized commands.

Hayashi is specifically accused of creating a virus similar to ransomware, which destroys computer data and demands ransom in cryptocurrency, using his home computer and smartphone on March 31, 2023. He has reportedly admitted to the allegations, telling police, “I thought I could do anything by asking AI. I wanted to make easy money.”

According to the MPD, the suspect is a former factory worker who had never studied IT professionally. He apparently told investigators that he became interested in generative AI through the news and “came up with the idea of creating ransomware.”

Hayashi is believed to have created the virus by combining design information provided in answers from multiple generative AIs. No damage by the virus has been confirmed.

Generative AIs such as OpenAI’s ChatGPT have an anti-abuse feature called “guardrails,” which refuses to answer questions about the creation of malicious programs. However, guardrails are not set on some of the AI models available on the internet by unknown creators.

It is believed that the suspect used unknown creators’ software and repeatedly asked indirect questions to conceal his intent, circumventing guardrails and getting the AIs to provide the information.

Hayashi was arrested in March of this year on suspicion of fraud and other charges for allegedly signing the contract for a SIM card using a forged ID. At that time, a homemade virus was found on his home computer and other devices.

Major IT companies in the United States have introduced self-regulation rules regarding generative AIs. In the European Union, too, measures against misuse such as the Artificial Intelligence Act have been enacted. In Japan, the government’s AI Strategy Council decided in May to begin studying laws and regulations targeting large-scale AI developers.

(Japanese original by Shohei Kato, Tokyo City News Department)