~/tag/agentic-ai
3 articles tagged agentic ai
The serious attack surface is no longer the chatbot. It is the agent's toolbelt: retrieval, memory, skills, CI pipelines, credentials, and every renderer or proxy trusted to move data. A field analysis of EchoLeak, malicious agent skills, coding-agent compromise, state-scale campaigns, and data poisoning, separating confirmed incidents from research demonstrations, with the containment architecture that actually holds up.
Muse, Amazon Quick, Hermes Agent and OpenClaw all promise to do more than chat. That is roughly where the similarities end. A practical comparison of four very different bets (hosted convenience, governed workplace AI, a self-improving runtime, and a local-first gateway), plus a security lens on each: where the lethal trifecta lives, and how to harden the two you run yourself.
Black Hat and DEF CON land in the same week in Las Vegas every year, and every year one story ends up getting told in the hallways more than the others. Usually it takes a couple of days to sort out which story won. This year, it took about a morning.