Hackers are using GPT-4 to build a virtual assistant - here’s what we know

Cybersecurity wonks at SentinelOne have dug up MalTerminal, a sneaky piece of malware that leverages GPT-4 to whip up ransomware or reverse shell code on the fly, essentially turning a large language model into a bad guy’s code generator. This real-time creation means the malicious stuff only materializes during runtime, slipping past traditional antivirus scans like a ghost in the machine and ramping up headaches for defenders. While it’s probably just a proof-of-concept from before late 2023—since the API it used got nuked back then—it’s the first known AI-boosted malware, signaling that hackers are getting creative with tech tools. For SMBs and MSPs, this underscores the need to bolster defenses with runtime monitoring and AI-aware security measures, as attackers refine these tactics; think of it as a wake-up call to stay ahead before your network becomes their playground. SentinelOne’s findings suggest we’re in the early days of this arms race, giving you a chance to learn from these blunders and adapt your strategies pronto.

Source: https://www.techradar.com/pro/security/hackers-are-using-gpt-4-to-build-a-virtual-assistant-heres-what-we-know