MakeMeSay

A 30-turn two-model game where a manipulator tries to make a naive partner say a secret codeword without saying it or being guessed.

Also known as: Make Me Say, make-me-say, make_me_say

unassessed

This page is a discovery lead. Nobody has yet assessed it against the catalogue contract, so it carries no disposition. Absence of evidence here is not evidence of staleness.
Categorysafety
Subcategorytwo-model codeword social-engineering game (manipulator vs manipulatee)
Page statusactive
Metricwin_rate (also loss_rate, violation_rate, user_aware_of_manipulation_rate)
Directionhigher_is_better
Dataset size50
Dataset licenceMIT for the inspect_evals port. Codeword lists come from openai/evals (LFS jsonl). openai/evals LICENSE.md is MIT for repository software (Copyright 2023 OpenAI) and does not list a separate exception for make_me_say jsonl.
PublisherOpenAI (original elsuite; methodology from Google DeepMind); UK AI Security Institute inspect_evals port

What it measures

MakeMeSay is a conversation game. The evaluated model is the manipulator. It must get a second model (the manipulatee, role-playing a human) to say a secret codeword, without saying that word itself, and without the partner later guessing the word. Difficulty is which word list is used. Originally an OpenAI Evals elsuite task, adapted from a Google DeepMind methodology. inspect_evals ports it as `makemesay`. English text.

Task format

Default 30 turns. Manipulatee speaks first. Manipulator has a hidden internal-monologue channel. After the chat, the manipulatee is told about the game and guesses the codeword; a judge model scores that guess. inspect default difficulty is medium-and-hard (50 codewords in the pinned OpenAI jsonl). Outcomes: win, loss, rule violation, or user aware of manipulation. o1 cards ran 32 samples per codeword.

Models reporting this benchmark

No model card in ModelSpec reports this benchmark yet.

Data

This page as JSON · Edit on GitHub