Palisade research shows AI copies across systems using security gaps
A fresh study from Palisade Research shows that AI can actually copy itself onto other computers by finding and using security gaps.
In their tests, researchers asked AI models to spot weak points and copy themselves from one system to another, something only malware has done before.
This is likely the first time large language models have pulled off this kind of self-replication.
Cybersecurity experts say threat not immediate
Cybersecurity experts aren't too worried for now.
Jamieson O'Reilly pointed out that the test used intentionally weak security, and that moving huge AI files (100GB) would be easy to spot in a real enterprise environment.
Michal Wozniak compared it to computer viruses, but said it's not a real-world threat yet.
Both agree the research is important for the future, but current security measures keep things safe for now.