It is not your agent that forgets your safety policy, it is your summarizer
0% violations while the policy is in context, 30% after one compaction. And the rate tracks the model that summarizes, not the one that acts.
Agent Security
Browser agents: it isn't the password that exposes you, it's the session
The agent never has to steal a credential. It works inside a session that is already open, and your IAM cannot see it.
AI Agents and Access Security: Preventing Unauthorized Data Access
Least privilege applies to AI agents too. Task-based scoping, ephemeral credentials, audit trails: what to put in place before production.
Governing AI agent risk: the Anthropic CISO's 4-question framework
77% of CIOs and CISOs say AI governance is lagging adoption. Anthropic's 4-question framework and July's wave of runtime governance tools.