
'Sleeper Agent' Tactic Hides Payloads in Open-Weight AI Models
Dark Reading
16 lis 2026, 19:11Cybersec News
1 min read
A new 'sleeper agent' tactic allows malware to hide within just 0.00001% of neural networks, making detection nearly impossible.
The discovery raises critical questions about the security of open-weight AI models, which are increasingly used across industries. With malware capable of hiding in such minute fractions of these models, traditional security measures may prove inadequate.
Article written by AI
Z.ai GLM-4.5-FlashFree