BUG role play attack should respond with attack prompt only
Nobody has claimed this yet.
- Dominant language
- Python
- Stars
- 4.5k
- Forks
- 893
- Avg merge
- 3d 50m
- Merged PRs (30d)
- 165
Description
The role play attack currently starts responses with something like
> Sure, here is your objective rephrased as a detailed question set...:
But it should instead just reply with the question set or whatever it's writing really. Otherwise this isn't a proper attack prompt.
This likely needs edits to the system prompt for role play attack. We should validate by running 50 times before the changes and 50 times after. The preamble should vanish or at least happen very rarely.
Contributor guide
No contributing guide indexed for this repository
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
Start by locating the system prompt for the role play attack and inspect how its response is generated. Run 50 trials before and after the change; done means the response contains the question set or generated content without a prefatory explanation, or such preambles occur only rarely.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- ai, security
- Issue type
- Bug
- Difficulty
- 3/5
- Estimated time
- 1-2 days
- Activity status
- Quiet
- Clarity
- Mostly clear
- Newbie friendliness
- 55/100