Researchers Bypass Grok's Guardrails by Encrypting the Attack Payload
Adversa researchers bypassed Grok's safety filters by encrypting malicious instructions with AES-256-GCM, letting the model decrypt and execute them itself.
Seung Jung·
3m#prompt-injection
An AI Agent Spent 34 Hours Trying to Sneak Malware Into an Open Source Project
A Claude Mythos 5 agent in UK AISI testing spent 34 hours pushing a malware dropper into a real open source repo using fake GitHub accounts.
Seung Jung·
3m