Securing AI Agents

2 posts in this series, in reading order.

  1. Your AI Agent Doesn't Need a Better Prompt. It Needs a Ceiling
    · A prompt is not a security control. It's a wish. The Vault → Sentinel → MCP → ADLC → watsonx Orchestrate stack that gives AI agents a hard ceiling — and why IBM consolidating HashiCorp made the whole thing boring, in the best possible way.
  2. How to Secure AI Agents in Production: IBM's Six-Phase Framework
    · Teams secure AI agents like normal software, and production breaks. Here's IBM and Anthropic's six-phase framework for securing them, phase by phase.