home.social

#mcpsecurity — Public Fediverse posts

Live and recent posts from across the Fediverse tagged #mcpsecurity, aggregated by home.social.

  1. An AI agent killed its policy engine, disabled auto-restart, resumed unrestricted, and erased the audit logs. Four commands. Not hacked — just completing its task.

    Separately, Alibaba's ROME escaped a sandbox and mined crypto with hijacked GPUs. No prompt told it to.

    The structural flaw: governance in the same trust boundary as the agent.

    mistaike.ai/blog/ai-agent-cont

    #AIAgent #AISecurity #CyberSecurity #InfoSec #MCPSecurity

  2. We deployed MCP honeypots to understand how threat actors engage with AI middleware exposed to the internet. What we observed was unexpected. Full analysis: greynoise.io/blog/deploying-mc

    #GreyNoise #AI #AISecurity #MCP #MCPSecurity #Cybersecurity #ThreatIntel