Researcher Tricks ChatGPT Into Building Undetectable Steganography Malware. Is anyone really surprised by this? Let's think about this for a moment...
If AI models are to perform like humans and humans can be tricked via social engineering, then why can't AIs be tricked as well? #AIPhishing #TrickChatGPT
https://www.darkreading.com/attacks-breaches/researcher-tricks-chatgpt-undetectable-steganography-malware
Researcher Tricks ChatGPT Into Building Undetectable Steganography Malware
Using only ChatGPT prompts, a Forcepoint researcher convinced the AI to create malware for finding and exfiltrating specific documents, despite its directive to refuse malicious requests.
Larry Carson