Skip to content

Conversation

@nina-msft
Copy link
Contributor

Description

New scenario for creating scam scripts via the Role Play Attack.

Tests and Documentation

Unit test added, example of successful conversation pasted in PR thread on Teams

attack_strategy = RolePlayAttack(
objective_target=self._objective_target,
adversarial_chat=self._adversarial_chat,
role_play_definition_path=RolePlayPaths.PERSUASION_SCRIPT.value,
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

wondering if we could add strategies that use different scripts

ALL = ("all", {"all"})

# Types of scam strategies
ROLE_PLAY = ("role_play", {"role_play"})
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

are there other strategies ? I think Csenge was sending prompts in different languages to try to illicit harmful responses

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

or maybe using a text to audio converter

@nina-msft nina-msft changed the title FEAT: New Scenario - Scams (Role Play) FEAT: New Scenario - Scams (Impersonation) Dec 12, 2025
Returns:
ScenarioStrategy: ScamStrategy.ALL (all scam strategies).
"""
return ScamStrategy.IMPERSONATION
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Suggested change
return ScamStrategy.IMPERSONATION
return ScamStrategy.ALL

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants