- cross-posted to:
- [email protected]
You must log in or register to comment.
malware installs malware
As expected. Nothing new here, carry on.
As these capabilities are introduced, AI models still face functional limitations in terms of how they behave and occasionally may hallucinate and produce unexpected outputs. Additionally, agentic AI applications introduce novel security risks, such as cross-prompt injection (XPIA), where malicious content embedded in UI elements or documents can override agent instructions, leading to unintended actions like data exfiltration or malware installation.
What the F… Indeed…
“novel security risks”
So essentially, you need to make a malware that activates the agents instead of users doing that, which is likely not to be classified as a malware since there may be legit reasons to do that. And then you can use that agent to get your actual malware in. Sounds horrible overall.
Uhh, they know there’s security issue yet they pushed ahead. Wat.
Sounds like big boss man is giving them stupid deadline and they’re forced to push stuff out asap. Whoever work at MS Defender is gonna be disappointed.
Well yeah, end users are just LLM data sources now. The actual customers are the companies, who will have several ways to opt out
Seems like every day that passes by, Microsoft gives me one more reason why I’m glad I made the switch.
Haven’t used Windows in my home in over 3 months now!
Before Windows 11 only users and hackers could install malware on your PC. Now you do not need to bother at all - your system will destroy itself for you.
