# Comprehension Audits to Mitigate Risks from Automated AI Research

arXiv: https://arxiv.org/abs/2610.10064 · HTML: https://mlsanity.com/paper/2610.10064/

## Key facts

- No citing papers in the map yet (data through 7 Oct 2026).
- Builds on 22 of its 89 references that are in the map.
- First posted on arXiv 7 Oct 2026.

## Builds on (most cited first, 22 of 22)

- [Concrete Problems in AI Safety](https://mlsanity.com/paper/1606.06565/) (2016)
- [Self-critiquing models for assisting human evaluators](https://mlsanity.com/paper/2206.05802/) (2022)
- [Monitoring Reasoning Models for Misbehavior and the Risks of Promoting Obfuscation](https://mlsanity.com/paper/2503.11926/) (2025)
- [Frontier Models are Capable of In-context Scheming](https://mlsanity.com/paper/2412.04984/) (2024)
- [Measuring Progress on Scalable Oversight for Large Language Models](https://mlsanity.com/paper/2211.03540/) (2022)
- [AI Control: Improving Safety Despite Intentional Subversion](https://mlsanity.com/paper/2312.06942/) (2023)
- [Toward Trustworthy AI Development: Mechanisms for Supporting Verifiable Claims](https://mlsanity.com/paper/2004.07213/) (2020)
- [Supervising strong learners by amplifying weak experts](https://mlsanity.com/paper/1810.08575/) (2018)
- [Your Brain on ChatGPT: Accumulation of Cognitive Debt when Using an AI Assistant for Essay Writing Task](https://mlsanity.com/paper/2506.08872/) (2025)
- [Safety Cases: How to Justify the Safety of Advanced AI Systems](https://mlsanity.com/paper/2403.10462/) (2024)
- [A sketch of an AI control safety case](https://mlsanity.com/paper/2501.17315/) (2025)
- [Let It Flow: Agentic Crafting on Rock and Roll, Building the ROME Model within an Open Agentic Learning Ecosystem](https://mlsanity.com/paper/2512.24873/) (2025)
- [How to evaluate control measures for LLM agents? A trajectory from today to superintelligence](https://mlsanity.com/paper/2504.05259/) (2025)
- [How AI Impacts Skill Formation](https://mlsanity.com/paper/2601.20245/) (2026)
- [Coordinated pausing: An evaluation-based coordination scheme for frontier AI developers](https://mlsanity.com/paper/2310.00374/) (2023)
- [Frontier AI Auditing: Toward Rigorous Third-Party Assessment of Safety and Security Practices at Leading AI Companies](https://mlsanity.com/paper/2601.11699/) (2026)
- [Measuring AI R&D Automation](https://mlsanity.com/paper/2603.03992/) (2026)
- [Position: Human-Centric AI Requires a Minimum Viable Level of Human Understanding](https://mlsanity.com/paper/2602.00854/) (2026)
- [A Framework for the Private Governance of Frontier Artificial Intelligence](https://mlsanity.com/paper/2504.11501/) (2025)
- [AI Researchers' Views on Automating AI R&D and Intelligence Explosions](https://mlsanity.com/paper/2603.03338/) (2026)
- [AI Agents Push Humans Out of the Loop](https://mlsanity.com/paper/2608.23642/) (2026)
- [Beyond the 'Diff': Addressing Agentic Entropy in Agentic Software Development](https://mlsanity.com/paper/2604.16323/) (2026)

## Abstract (arXiv)

AI is already writing a majority of code for frontier AI labs. This creates a safety risk if there is insufficient human oversight. Existing work proposes minimum comprehension thresholds and unaided checks to mitigate this. To our knowledge, however, there is currently no published frontier-AI assurance regime that requires demonstrated evidence that the responsible humans understand what they are building as a precommitted condition for continuing development or usage. We propose comprehension audits, a novel development-process assurance mechanism in which the responsible people explain R&D contributions to auditors to demonstrate understanding. With independent administration and graded reports, they provide a gate: development of a contribution stops based on a failure to demonstrate human understanding until remediated, with escalating consequences for repeated failures. Our analysis of leading open-source AI projects finds increased output of code with reduced human review commentary rates per line of code, with far lower rates for automated fleet accounts. We advocate for labs to conduct them with embedded independent auditors.