Notifications
Clear all
Topic starter
01/08/2026 10:44 pm
TL;DR: Post-training can improve offensive security agents by combining supervised fine-tuning, reinforcement learning from verified rewards, live web apps, browser control, and asynchronous rollout training, according to Novee. The practical implication is that AI security teams now have to govern model behaviour, tool use, and training pipelines as a coupled control surface, not as separate concerns.
NHIMG editorial — based on content published by Novee: Gym-RL, a post-training pipeline for offensive security
By the numbers:
- With tuned Docker and machine settings we comfortably run 500–1,000 concurrent rollouts on a single 8×B200 node.
Questions worth separating out
Q: How should security teams govern AI agents that choose tools at runtime?
A: Security teams should treat runtime tool choice as a governed access event, not a normal application call.
Q: Why do AI agent training pipelines need identity-style controls?
A: Because the pipeline grants temporary access to models, browsers, apps, and data stores, which is functionally an identity problem.
Q: What do security teams get wrong about reward-based model training?
A: They often assume that a successful-looking output proves the right behaviour was learned.
Practitioner guidance
- Define AI agent access boundaries Map every agent rollout to the exact app, browser, model gateway, and sandbox permissions it needs.
- Treat validators as controlled security assets Subject deterministic validators to the same change control, test coverage, and review discipline you would apply to a production detection rule.
- Separate rollout telemetry from training data Keep tokens, traces, and logprobs in a governed pipeline with explicit retention and access policies.
What's in the full article
Novee’s full post covers the operational detail this post intentionally leaves for the source:
- The rollout orchestration design for live app instances, browser sessions, and sandbox lifecycles.
- The model gateway mechanics for trace capture, sticky routing, and tokens-in, tokens-out handling.
- The validation framework used to turn exploit execution into a reward signal.
- The concurrency and infrastructure details behind thousands of app instances and hundreds of parallel rollouts.
👉 Read Novee’s full analysis of NoveeGym-RL and post-training for offensive security →
Agentic pentesting post-training: what it means for AI security?
Explore further
01/08/2026 10:53 pm
Post-training is becoming an identity and governance problem, not only a model-quality problem. Once an AI system can browse, act, and learn from tool use, its permissions, traces, and reward paths behave like security controls. That places agent identity, model identity, and environment identity in the same governance frame, especially where NHI-style credentials or ephemeral access tokens are used. Practitioners should read this as a control-plane issue, not a training-only issue.
A question worth separating out:
Q: What should organisations do before scaling agentic pentesting systems?
A: They should verify sandbox isolation, credential scope, telemetry retention, and offboarding for every rollout path. Scaling should come after those controls are tested under load, because concurrency multiplies any weakness in resource lifecycle management. The goal is to make each episode disposable, auditable, and impossible to overreach.
👉 Read our full editorial: Post-training agentic pentesting models changes AI security risk
