https://openaiguardrails.org/blog/ Research and engineering notes on AI guardrails, policy-as-code, and safe LLM deployment. en-us Sun, 10 May 2026 18:55:54 GMT https://openaiguardrails.org/blog/the-forge-hardening-ai-safety-in-the-noir-sandbox.html https://openaiguardrails.org/blog/the-forge-hardening-ai-safety-in-the-noir-sandbox.html Sun, 10 May 2026 00:00:00 GMT The Forge is Noir’s adversarial AI safety sandbox for building, testing, and deploying production-grade guardrails. Replay attacks, validate rail logic, simulate prompt injection scenarios, and promote hardened policies directly into CI/CD and distributed enforcement pipelines. noreply@openaiguardrails.org (auth0|69ebedaa1d4731384e69df25) DevSecOps Noir Guardrails Policy Enforcement https://openaiguardrails.org/blog/probe-terminal-agentic-ai-attack-surface.html https://openaiguardrails.org/blog/probe-terminal-agentic-ai-attack-surface.html Sun, 10 May 2026 00:00:00 GMT Agentic AI introduces a new security frontier where prompts, orchestration logic, and runtime behavior become the attack surface. The Probe Terminal enables automated adversarial testing, real-time risk grading, and continuous AI security enforcement without sacrificing performance. noreply@openaiguardrails.org (auth0|69ebedaa1d4731384e69df25) Agentic AI AI Security Prompt Injection LLM Security