# Harmonizing AI Safety Thresholds

arXiv: https://arxiv.org/abs/2607.16112 · HTML: https://mlsanity.com/paper/2607.16112/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 15 of its 36 references that are in the map.
- First posted on arXiv 17 Jul 2026.

## Builds on (most cited first, 15 of 15)

- [Cybench: A Framework for Evaluating Cybersecurity Capabilities and Risks of Language Models](https://mlsanity.com/paper/2408.08926/) (2024)
- [CyberGym: Evaluating AI Agents' Real-World Cybersecurity Capabilities at Scale](https://mlsanity.com/paper/2506.02548/) (2025)
- [CRISPR-GPT for Agentic Automation of Gene-editing Experiments](https://mlsanity.com/paper/2404.18021/) (2024)
- [Artificial intelligence and biological misuse: Differentiating risks of language models and biological design tools](https://mlsanity.com/paper/2306.13952/) (2023)
- [A Rosetta Stone for AI Benchmarks](https://mlsanity.com/paper/2512.00193/) (2025)
- [Measuring AI Agents' Progress on Multi-Step Cyber Attack Scenarios](https://mlsanity.com/paper/2603.11214/) (2026)
- [Risk thresholds for frontier AI](https://mlsanity.com/paper/2406.14713/) (2024)
- [LabOS: The AI-XR Co-Scientist That Sees and Works With Humans](https://mlsanity.com/paper/2510.14861/) (2025)
- [Contemporary AI foundation models increase biological weapons risk](https://mlsanity.com/paper/2506.13798/) (2025)
- [Generative AI for Biosciences: Emerging Threats and Roadmap to Biosecurity](https://mlsanity.com/paper/2510.15975/) (2025)
- [Toward Quantitative Modeling of Cybersecurity Risks Due to AI Misuse](https://mlsanity.com/paper/2512.08864/) (2025)
- [A Methodology for Quantitative AI Risk Modeling](https://mlsanity.com/paper/2512.08844/) (2025)
- [Global Cybercrime Damages: A Baseline for Frontier AI Risk Assessment](https://mlsanity.com/paper/2603.20570/) (2026)
- [TitanCA: Lessons from Orchestrating LLM Agents to Discover 100+ CVEs](https://mlsanity.com/paper/2604.17860/) (2026)
- [Estimating the Social Cost of Corporate Data Breaches](https://mlsanity.com/paper/2603.21270/) (2026)

## Abstract (arXiv)

Frontier AI companies have published capability thresholds that differ substantially, making it difficult for third parties to verify whether a threshold has been crossed or to compare requirements across companies. Moreover, without common minimum thresholds, risk mitigation may be inconsistent, creating a potential race to the bottom in safety standards. We develop a methodology for deriving harmonized thresholds across three risk domains. For misuse risks (cyber and biological), we take expected harm as the key primitive and use an explicit risk-modeling approach that accounts for risk channels and model release conditions. For automated AI R&D, we base our proposed threshold on the observed rate of AI progress rather than expected harm. Our analysis expands upon prior work and highlights existing empirical gaps and limitations.