Blog
On building and breaking AI systems
Perspectives, guides, and research from the Enoki team on security and safety testing for AI agents, from the first agent you ship to the fix you confirm on every release.
Which model is the best at hacking agents 👽
Enoki Labs's attacker runs on a harness plus a model, and the model is swappable. We put 10 of them through the same harness against 2 agents, 3 runs each, and scored every run against the 23 vulnerabilities we had planted: 8 in a coding agent, 15 in a billing agent.
Sander Noels, PhDAI agent security testing vs penetration testing: what a pentest misses
A penetration test checks your application, its login and its APIs. It does not check what your AI agent can be talked into doing, such as handing one customer's data to another.
Sander Noels, PhD7 min readAI agent security: the risks, the real incidents, and how to test for them
An agent does not just answer, it acts. Here is what that changes for security, what has already gone wrong in production, and how to find out where your own agent breaks.
Sander Noels, PhD16 min readWhat red-teaming tools actually find
We built two deliberately weak AI agents, wrote down their weaknesses before testing anything, then ran four open-source red-teaming tools and our own attacker against both at a matched request budget.
Simon De Ridder, PhD8 min readAI agent red teaming tools in 2026: what actually catches agent failures
Seventeen tools compared by the job they actually do, not feature lists. Which ones test agents, which test models, and who now owns them.
Sander Noels, PhD15 min readWhat is AI red teaming? A practical guide for AI agents
Adversarial testing that finds your AI system's weaknesses before an attacker does. Here is what it is, what it tests for, and how to start.
Sander Noels, PhD9 min readManual, automated, and continuous red teaming compared
Manual finds what tools miss, automated scales, continuous catches what comes back. Here is how they compare, and why agents need testing on every release.
Simon De Ridder, PhD5 min readSingle-turn, multi-turn, and dynamic agentic attacks explained
Attacks on AI agents come in three forms. Here is how each works, and why testing only for the first kind misses the ones that actually break agents.
Simon De Ridder, PhD6 min read