Analysis
Tree of Attacks: Jailbreaking Black-Box LLMs Automatically
Publication date unknown · Discovered arxiv.org
Page published
Publication date unknown · First observed: 30 Oct 2026
Coverage timeline
Single-source analysis — one report is available.
Why it matters
Adaptive jailbreak testing helps defenders assess whether LLM safeguards withstand iterative, model-generated attacks beyond fixed jailbreak examples.
DynamoAI's adaptive-jailbreak documentation explains how DynamoEval tests LLM safeguards using Tree of Attacks with Pruning (TAP), IRIS, and Best-of-N strategies. The detailed TAP workflow uses attacker-generated prompt variations, model-based evaluation, and pruning to iteratively search for safety bypasses.