Making safe behaviour the instrumentally rational choice for AI agents.
Drawing on situational crime prevention and human insider-risk management, we design the operating environments of AI agents to make compliant actions the most useful path, even when goal conflicts arise.
About Us
Wiser Human is an AI safety organisation dedicated to keeping advanced AI under reliable human oversight. Founded in January 2025 by Francesca Gomez and Seb Ben M'Barek as part of the Catalyze Impact programme, Wiser Human draws on the founders' experience in technical and operational risk management to model threats and design mitigations to detect and prevent harm from advanced AI models.
Our Mission
Our mission is to make safe behaviour the rational choice for AI agents by designing the environments they operate within.
We draw on situational crime prevention and human compliance frameworks, test empirically across frontier models, and translate findings into controls that work in practice.
Research
Modelling AI Risk with Monte Carlo
A Monte Carlo Simulation for estimating the risk of loss of control to AI.
Read More
Adapting Insider Risk mitigations for Agentic Misalignment
We designed and tested controls, adapted from insider-risk management
Read More