Built Among Us Sim to Test LLM Deception: Claude vs. Kimi
Aryan100100 · x · 2026-08-22
Inspired by the Generative Agents Simulacra paper, the author built simulations to observe how LLMs play games against each other.
- Testbed: Chose "Among Us" (Mafia style game) and mapped every pixel to build a custom sim (pinned on profile).
- Matchup: Had Claude play against Kimi to gather data on LLM deception in game scenarios.
- Duration: Working on variations of this idea since October 2025 (10 months).
More from coding & agent
- Chroma Launches Foundation for Self-Improving Agent Memory — nptacek · 2026-08-22
- OJO AI Design Agent Tested: Generating Full Landing Pages with One Prompt — PrajwalTomar_ · 2026-08-22
- Natural Language Programming: Using AI Agent as a runtime for Markdown-defined apps — holy_serp · 2026-08-22
- 5 prompts to turn Claude into a free wireframe-to-code studio — nikola_mr64990 · 2026-08-22
- Coding agents need less freedom, not more: The risk of unchecked autonomy — phucphungbk · 2026-08-22
- Guide: Building a Browser Harness/MCP for AI Agents vs Playwright — Jin-109 · 2026-08-22