Lessons from External Review of DeepMind's Scheming Inability Safety Case
For AI developers and regulators, it demonstrates how external review can identify flaws in safety cases that internal biases may miss.
The paper performs an external review of DeepMind's safety case for scheming inability, uncovering new concerns that affect its scope and decision-making applicability, and provides recommendations for future external reviews.
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.