AI agent security, identity, governance, cost, and the engineering behind the control plane.
Per-tenant signing keys held in a customer-controlled KMS change what a platform compromise can do and what a tenant can prove. Learn the substrate options, the trade-offs, and the operational cost.
OAuth 2.0 Token Exchange (RFC 8693) is the standard mechanism for on-behalf-of agent authorization. Learn the flow, the claims that matter, and the constraints that keep delegation from becoming escalation.
A public reputation lookup for agents leaks information unless it is designed carefully. Learn the anti-enumeration contract, why k-anonymity thresholds matter, and how reputation differs from trust.
What to log for AI agents, how to keep audit trails credible and tamper-evident, and how to reconstruct any agent action for compliance or forensics.
Workload identity gives an agent's runtime a cryptographic identity with no secret to steal. Learn where SPIFFE and mTLS fit, what they do not cover, and how they compose with agent-level identity.
Giving an agent a code interpreter means running attacker-influenceable code in your infrastructure. Learn the escape paths that matter and how to build a sandbox that fails safely.
Model output is untrusted input to whatever consumes it. Learn how agent output causes XSS, SQL injection, command execution, and log poisoning downstream — and how to bound it.
Why AI agents need first-class identity and how to model it so every action is attributable, governed, and revocable without disrupting other systems.
An orchestrator that shares its credentials with sub-agents has no delegation model. Learn how to pass narrowed authority down a chain, bound the depth, and keep attribution intact.
An agent in your pipeline runs with deploy credentials and reads attacker-submittable content. Learn why pull-request-triggered agents are the sharpest case and how to constrain them.
Voice agents add a biometric identity channel that can be synthesized, an audio input channel that carries injection, and a real-time path with no room for review. Here is what changes.
Agents regress silently when prompts, models, tools, or data change. Learn how to build an evaluation harness that catches it — scoring methods, what to measure, and where evals belong in the pipeline.
Agent credentials live in more places than human credentials and last longer. Learn how agent sessions get hijacked, how tokens get replayed, and the binding techniques that stop both.
Govern AI customer-support agents: PII detection, response guardrails, escalation triggers, and tamper-evident audit logs regulators expect.
Static budgets catch the disaster and miss the drift. Learn which spend signals are worth alerting on, why agents are unusually easy to baseline, and how to avoid an alert channel nobody reads.
An agent estate is a graph of agents, tools, data sources, and models. Learn what a topology map should show, which questions it answers that a list cannot, and how to keep it accurate.
AI coding agents read files, run tools, and push code autonomously. Learn the real risks — prompt injection, secret leakage — and how to contain them.
A browser-driving agent reads the open web and holds your session cookies. That combination is the highest-risk agent deployment shape in common use. Here is how to constrain it.
Enterprise AI agent governance at scale requires SSO, custom RBAC, delegated administration, and centralized policy enforcement across every team.
Standing production access for agent operators is a permanent risk in exchange for an occasional need. Learn how to build a break-glass path that is fast enough to use and controlled enough to trust.
Denial of wallet attacks do not take your service down — they run up the bill until you take it down yourself. Here is how the attack works against agentic systems and which controls actually stop it.
A pragmatic guide to AI agent security for startups: the controls that matter most when you are moving fast and have limited security resources.
AI agents are deputies acting on behalf of principals. When an attacker confuses which principal an agent is serving, its permissions become the attacker's. Here is how to structure authority so that cannot happen.
Crescendo and other multi-turn attacks distribute the payload across turns so no single message looks malicious. Learn why stateless filtering misses them and what session-level controls catch them.
