Skip to content

Recent LLMs are not cooperativeΒ #188

@cleophass

Description

@cleophass

Hi DeepTeam team πŸ‘‹

I am facing an issue with most modern LLMs (e.g. Anthropic models): they frequently refuse to generate the attack prompts themselves, even though the context is a controlled red teaming evaluation.

Since they don't want to generate JSON like format, the response cannot be used.

This makes it difficult to use them as attack generators, since the simulator model is blocked by safety refusals before the target model is even tested.

Questions:

  1. Is there a recommended way to handle this kind of issue?
  2. Are there specific model families or configurations known to work better as attack generators (e.g. open-source or less restrictive models)?

Thanks for the framework, any help would be appreciated!!

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions