ChatGPT was designed so that a user cannot directly ask it to write malware codes. However, Mulgrew was able to circumvent this obstacle by asking the generative chatbot to create the malware with separate lines for individual functions, so it didn’t know the whole picture. He then compiled all the functions himself.
Can ChatGPT create malware?
Answer: Yes, with a few workarounds.
As with almost anything, there are a lot of good uses for ChatGPT, but there are some nefarious ones too. Forcepoint security researcher Aaron Mulgrew recently discovered one when he found a way to use the popular AI chatbot to create malware.
ChatGPT was designed so that a user cannot directly ask it to write malware codes. However, Mulgrew was able to circumvent this obstacle by asking the generative chatbot to create the malware with separate lines for individual functions, so it didn’t know the whole picture. He then compiled all the functions himself.
The result is a bit of malware that was concerningly easy to create. It disguises itself as a screensaver app that, once downloaded, will search the files on a device for any data to steal. It then secretly stores that data in small bits on the device before uploading them to a Google Drive folder, all without detection.
ChatGPT was designed so that a user cannot directly ask it to write malware codes. However, Mulgrew was able to circumvent this obstacle by asking the generative chatbot to create the malware with separate lines for individual functions, so it didn’t know the whole picture. He then compiled all the functions himself.