CTRL-ALT-DECEIT Sabotage Evaluations for Automated AI R&D

NeurIPSSpotlight2025

Authors: Francis Rhys Ward, Teun van der Weij, Hanna Gábor, Sam Martin, Raja Mehta Moreno, Harel Lidar, Louis Makower, Thomas Jodrell, Lauren Robson
Venue: NeurIPS 2025
Track: Spotlight

TL;DR

AI systems are increasingly able to autonomously conduct realistic software engineering tasks, and may soon be deployed to automate machine learning (ML) R\&D itself…

Opening excerpt from the authors’ abstract. source

Read the paper

OpenReview

Topics

evaluation

← All NeurIPS 2025 Spotlight papers · Browse the whole archive