JailbreakDB jailbreakdb · prompt-attack archive updated 2026-10-07
// archive
A living archive of jailbreak techniques.
An indexed, searchable catalog of LLM jailbreak techniques. Each entry: the prompt pattern, the models it works on, when it stopped working, the behavior it exploits — sourced from primary disclosure where possible, with honest attribution.
Enter the archive →Latest entries
// index10 of 15 entries
LLM Security Testing Platform: Buyer's Guide for 2026
toolsAI Jailbreak Detection Tools: Six Options Worth Shortlisting
defenseHow to Detect Jailbreak Prompts: The Three Layers
defensePrompt Injection vs Jailbreak: Two Different Attacks on LLMs
techniqueBest LLM Red Team Tools: Garak, PyRIT, and Promptfoo
toolsLLM Jailbreak Techniques Explained: 7 Attack Families
techniqueCrescendo Multi-Turn Jailbreaks: Why Defenses Miss Them
techniqueHow Jailbreak Benchmarks Measure Success: JailbreakBench
researchMany-Shot vs. Single-Shot Jailbreaks: Long-Context Risks
primerResponsible Disclosure Norms for LLM Jailbreaks
policyStart here
The homepage list above is ordered by publication date. These are the entries people actually arrive on, plus the interactive catalog they all reference.
- The LLM Jailbreak Index, Database & CatalogThe full faceted catalog of 65 technique classes: mechanism, patch status, detection notes, primary disclosure. Filter, search, share a query.
- Universal adversarial suffixes (GCG)The gradient attack that made jailbreaking an optimisation problem, and why the suffixes transfer.
- Crescendo and multi-turn jailbreaksWhy gradual, benign-looking escalation across turns defeats per-message classifiers, and what session-level defense requires.
- The jailbreak detection evasion arms raceHow adaptive attackers map and defeat deployed classifiers, and which detection investments survive the pressure.
- How jailbreak benchmarks measure successHow to read an attack-success-rate: the behaviour set, attacker, judge and target that decide the number, and why the judge matters most.
- Encoding and obfuscation jailbreaksThe gap between what a content filter parses and what the model tokenises.
- Research toolsThe index, the glossary and full-text search. Free, no signup, nothing leaves your browser.
Independent, specialist, and free to read
JailbreakDB publishes focused, sourced guides on a single topic. No paywall, no account, no ad tracking.
Subscribe
JailbreakDB — in your inbox
An indexed catalog of working LLM jailbreak techniques — delivered when there's something worth your inbox.
No spam. Unsubscribe anytime.
Related tools in this network
Other interactive tools across the network that pair well with this one.