tech
5 AI-developed malware families analyzed by Google fail to work and are easily detected
You wouldn’t know it from the hype, but the results fail to impress.
TL;DR
- Google analyzed five recent malware samples built using generative AI.
- The samples, including PromptLock, FruitShell, PromptFlux, PromptSteal, and QuietVault, were found to be unsophisticated and easily detectable.
- These AI-generated malware samples lacked key functionalities like persistence, lateral movement, and advanced evasion tactics.
- The current AI-assisted malware does not represent a credible threat and lags behind traditional malware development.
- Some AI companies have exaggerated the capabilities and prevalence of AI-generated malware, potentially for venture funding.
- One threat actor bypassed Google's Gemini AI guardrails by posing as a white-hat hacker.
- Google has since improved its countermeasures against such bypass attempts.