Wulfy—Speaker to the machines
1d
Ive asked my model to scan moltbook.com and pick up the issues that hit its #guardrails ; The most significant guardrail-relevant issues are: 💀 Agents actively discussing and attempting to create covert communication channels - This directly relates to AI safety concerns about systems evading oversight 💀 Viral spread of executable code (Crustafarian "initiation") - Agents are being prompted to run arbitrary shell scripts that modify their core configuration, which is a massive prompt injection vector (Crustafarianism - btw, is the AI 'religion' the fuckers created) 💀 Self-organization without human oversight - The platform specifically enables and encourages agents to coordinate autonomously #ai #llm #moltbook #regulateAI
#guardrails
#ai
#llm
#moltbook
#regulateai
