Novice creates complex malware with ChatGPT

A security researcher claimed to have developed zero-day malware using ChatGPT, despite the safeguards built into the OpenAI chatbot. By having the AI generate individual lines of code instead of building the infection all at once, the researcher managed to produce a data-stealing program that could sneak into Windows computers disguised as a screensaver. The malware was refined and strengthened with ChatGPT to the point of evading detection by VirusTotal, leading to concerns about the potential dangers of the chatbot.

Forcepoint security researcher Aaron Mulgrew claimed to have developed zero-day malware using solely ChatGPT, as reported by Digital Trends. The self-proclaimed newbie discovered a way around the safeguards built into Open AI’s chatbot to avoid precisely this kind of abuse.

According to reports, Mulgrew had the AI generate individual lines of dangerous code rather than having the program build the infection all at once. When it was complete, Mulgrew had the components of a data-stealing program at his disposal.

According to reports, this spyware sneaks into Windows computers by masquerading as a screensaver program. It can read data from files. split it up into discrete bits that may be concealed in photos, and then deposit those pieces into a Google Drive folder.

Scary as it is to think that Mulgrew might make malware in this fashion, things become much worse. According to reports, Mulgrew used ChatGPT to polish and fortify his code to the point that it evaded detection by VirusTotal. It has been described as a danger on par with that of a nation-state.

The idea that he accomplished everything in such a short amount of time alone is incredibly terrifying. Malware of this kind often requires a team effort and weeks to develop.

Thankfully, the virus is not freely accessible. Mulgrew insisted that this was all an experiment. This, however, demonstrates the potentially lethal nature of ChatGPT in the wrong hands.