Loophole: Adversarial Agents To Stress Test Your Morality — Brendan Rappazzo, Morgan Stanley

Loophole, an open-source adversarial agent framework, enables users to codify personal moral principles into a structured legal system and stress-test them against synthetic case law. By utilizing LLMs to simulate adversarial agents, the system identifies contradictions, loopholes, or instances of overreach, prompting users to refine their moral definitions through an iterative, auto-patching process. Beyond personal moral exploration, the project extends into practical applications such as defining constitutions for AI chatbots, facilitating decentralized contracts, and simulating legislative voting behavior. By training models on U.S. Senator voting records or representative demographic personas, the framework allows for the simulation of legislative outcomes and the optimization of bill language to achieve broader consensus. This approach shifts the focus from abstract value-based arguments to nuanced, verifiable moral reasoning.