RockB editorial guides

All-AI-Jailbreaks Archive Review: What the Repository Can and Cannot Prove

All-AI-Jailbreaks is a collection of community prompt experiments. It can supply leads for defensive research, but a filename naming a model does not establish that the prompt works against that model today. This review examines the repository’s inventory and documentation, not the effectiveness of its payloads. The useful question is whether you can turn a claimed failure into a reproducible, authorized test of your own application. A collection size or a dramatic title is not an attack-success rate. ...

August 26, 2026 · 4 min · baeseokjae
LLM Red Teaming Guide: Scope, Failure Evidence and Regression Cases

LLM Red Teaming Guide: Scope, Failure Evidence and Regression Cases

An LLM red-team exercise is useful when it reveals a failure you can reproduce, explain and fix. Collecting dramatic prompts without an application boundary or an outcome definition produces little evidence about a deployed agent. This guide focuses on authorized testing of systems you own or have permission to assess. It provides a runbook and result schema, not a claimed security score for a model or product. Define the trust boundary OWASP’s prompt-injection description covers adversarial instructions entering through prompts or external content. Its prevention guidance recommends layered controls instead of treating natural-language instructions as an authorization system. ...

May 10, 2026 · 3 min · baeseokjae
OpenAI Acquires PromptFoo: What It Means for AI Security Testing in 2026

OpenAI Acquires PromptFoo: What It Means for AI Security Testing in 2026

OpenAI acquiring PromptFoo is not a talent grab — it is a strategic acknowledgment that AI security testing is no longer optional infrastructure. With 93% of organizations now shipping AI-generated code and only 12% applying equivalent security standards, the attack surface is enormous and growing. PromptFoo was the most mature open-source tool purpose-built for LLM red-teaming, and OpenAI buying it means the company is betting that security evaluation needs to be a first-class part of the developer workflow, not an afterthought bolted on by a third-party CLI. ...

May 10, 2026 · 13 min · baeseokjae