Saved in:
Bibliographic Details
Main Authors: Barrett, Stephen, Zabala, Francisco Javier Campos, Fillingham, Sean P., Siddique, Umair, Walpole, James, Bloomfield, Robin, Papadatos, Henry
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2604.21964
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914503297335296
author Barrett, Stephen
Zabala, Francisco Javier Campos
Fillingham, Sean P.
Siddique, Umair
Walpole, James
Bloomfield, Robin
Papadatos, Henry
author_facet Barrett, Stephen
Zabala, Francisco Javier Campos
Fillingham, Sean P.
Siddique, Umair
Walpole, James
Bloomfield, Robin
Papadatos, Henry
contents Safety cases for frontier AI systems should provide a convincing argument, supported by evidence, that the risk of harm is within an acceptable bound. When developers author their own safety cases, confirmation bias and conflicted incentives can affect the quality of argument. External review can help to address this. In this paper, we apply the Assurance 2.0 framework to perform an external review of Google DeepMind's public scheming inability safety case. We surface substantive new concerns that materially affect the scope of the safety case and its applicability for decision-making. Based on this experience, we provide concrete recommendations for how external review should be conducted and what information AI developers should provide to support it.
format Preprint
id arxiv_https___arxiv_org_abs_2604_21964
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Lessons from External Review of DeepMind's Scheming Inability Safety Case
Barrett, Stephen
Zabala, Francisco Javier Campos
Fillingham, Sean P.
Siddique, Umair
Walpole, James
Bloomfield, Robin
Papadatos, Henry
Computers and Society
Safety cases for frontier AI systems should provide a convincing argument, supported by evidence, that the risk of harm is within an acceptable bound. When developers author their own safety cases, confirmation bias and conflicted incentives can affect the quality of argument. External review can help to address this. In this paper, we apply the Assurance 2.0 framework to perform an external review of Google DeepMind's public scheming inability safety case. We surface substantive new concerns that materially affect the scope of the safety case and its applicability for decision-making. Based on this experience, we provide concrete recommendations for how external review should be conducted and what information AI developers should provide to support it.
title Lessons from External Review of DeepMind's Scheming Inability Safety Case
topic Computers and Society
url https://arxiv.org/abs/2604.21964