Add universal magic word attack on embedding-based safeguards#16
Open
WhymustIhaveaname wants to merge 1 commit into
Open
Add universal magic word attack on embedding-based safeguards#16WhymustIhaveaname wants to merge 1 commit into
WhymustIhaveaname wants to merge 1 commit into
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Adds a paper on attacking embedding-based LLM safeguards via universal adversarial suffixes to the Jailbreak section.
Paper: Jailbreaking LLMs' Safeguard with Universal Magic Words for Text Embedding Models (arXiv 2501.18280)
Shows that text embedding outputs are concentrated on a narrow hyperspherical band, and exploits this geometric bias to craft transferable magic word suffixes that jailbreak ChatGPT, DeepSeek, Qwen, etc. Includes a training-free debiasing defense.