LLM Jailbreaking Techniques

---
created:
  - 2024-11-27T08:53
modified: 2024-11-27 08:59
tags:
  - llm
  - large-language-model
  - nlp
  - natural-language-processing
  - security
  - cybersecurity
  - jail-break
  - red-team
  - red-teaming
  - vulnerability
type:
  - map-of-content
status:
  - ongoing
---
Main body of note goes here

| Technique                                         | Description | Link(s)                                   |
| ------------------------------------------------- | ----------- | ----------------------------------------- |
| Bijection Learning                                |             | https://arxiv.org/abs/2410.01294          |
| Multi-turn jailbreaks via Monte Carlo Tree Search |             |                                           |
| Transferring attacks from ACG                     |             | https://blog.haizelabs.com/posts/acg/     |
| Evolutionary Algorithms                           |             | https://github.com/haizelabs/dspy-redteam |
| BEAm search-based AdverSarial aTtack (BEAST)      |             | https://arxiv.org/abs/2402.15570          |
## References
* https://blog.haizelabs.com/
## Related
* Links to other notes which are directly related go here