We introduce scheming honeypot evaluations, a framework for testing whether models will pursue instrumental goals if given the opportunity. Our scheming honeypot evaluations take the form of coding tasks in Google's alignment research codebases. In a real internal deployment setting, Gemini models do not demonstrate unprompted scheming. If prompts explicitly encourage agency (situational awareness or goal-directedness) and/or give the model a hidden goal, models sometimes scheme or attempt sabotage. Validating the realism of our setting, models show low rates of evaluation awareness, usually due to agency prompts rather than the environments.
@article{arxiv.2605.29729,
title = {Realistic honeypot evaluations for scheming propensity},
author = {Victoria Krakovna and David Lindner and Lewis Ho and Sebastian Farquhar and Rohin Shah},
journal= {arXiv preprint arXiv:2605.29729},
year = {2026}
}