Arguments about AI and biological risk
Which located texts, from 2023 to 2026, state, contest or bear on the published arguments about AI and biological risk: whether current systems already help an actor seeking biological harm, where the binding barrier lies, what published evaluations establish, whether design tools are a distinct case, whether synthesis screening is the control point, whether open weights raise the risk, and whom assessment must cover? For each proposition, which text is the earliest located?
Each row is one text’s stance on one proposition, as a stored record states it: articulates, contests or bears on. A stance belongs to the text, not to its author. Rows are in date order, which records no influence between texts. Authors appear as the texts’ own bylines print them, pseudonyms included. “Earliest located” marks only a text the collection’s records name as the earliest the searches found.
R · What current systems and evaluations show
D1 Currently available AI systems increase an actor's biological-misuse capability
- Articulates · Earliest locatedarXiv (preprint, Emily H. Soice, Rafael Rocha, Kimberlee Cordova, Michael Specter, Kevin M. Esvelt)Can large language models democratize access to dual-use biotechnology?2023-06-06 · publishedInspect the record ↗
- ArticulatesarXiv (preprint, Jonas B. Sandbrink, Nuffield Department of Medicine, University of Oxford)Artificial intelligence and biological misuse: Differentiating risks of language models and biological design tools2023-06-24 · publishedInspect the record ↗
- Contests · Earliest locatedRAND CorporationThe Operational Risks of AI in Large-Scale Biological Attacks: Results of a Red-Team Study2024-01-25 · publishedInspect the record ↗
- ContestsInformation Technology and Innovation Foundation (blog; Martin Makaryan)The Shift in Rhetoric on AI and Biothreats Is a Lesson on the Risks of Premature Regulation2024-04-30 · publishedInspect the record ↗
- ContestsarXiv (preprint, Aidan Peppin, Anka Reuel, Stephen Casper and others)The Reality of AI and Biorisk2024-12-02 · publishedInspect the record ↗
- ArticulatesRAND Corporation (working paper WR-A3853-1), posted to arXivContemporary AI foundation models increase biological weapons risk2025-06-12 · publishedInspect the record ↗
D2 The binding barrier is physical execution rather than information
- Contests · Earliest locatedarXiv (preprint, Emily H. Soice, Rafael Rocha, Kimberlee Cordova, Michael Specter, Kevin M. Esvelt)Can large language models democratize access to dual-use biotechnology?2023-06-06 · publishedInspect the record ↗
- Bears onCenter for Security and Emerging Technology, Georgetown UniversityAI and Biorisk: An Explainer2023-12 · publishedInspect the record ↗
- ContestsRAND Corporation (working paper WR-A3853-1), posted to arXivContemporary AI foundation models increase biological weapons risk2025-06-12 · publishedInspect the record ↗
- ContestsAnthropic (Frontier Red Team)LLMs and biorisk2025-09-05 · publishedInspect the record ↗
D3 Published evaluations do not establish real-world biological risk
- Articulates · Earliest locatedarXiv (preprint, Aidan Peppin, Anka Reuel, Stephen Casper and others)The Reality of AI and Biorisk2024-12-02 · publishedInspect the record ↗
- ArticulatesRAND Corporation (working paper WR-A3853-1), posted to arXivContemporary AI foundation models increase biological weapons risk2025-06-12 · publishedInspect the record ↗
- ArticulatesAnthropic (Frontier Red Team)LLMs and biorisk2025-09-05 · publishedInspect the record ↗
G · What follows for governance
D4 Biological design tools pose risks distinct from those of language models
- Articulates · Earliest locatedarXiv (preprint, Jonas B. Sandbrink, Nuffield Department of Medicine, University of Oxford)Artificial intelligence and biological misuse: Differentiating risks of language models and biological design tools2023-06-24 · publishedInspect the record ↗
- ArticulatesarXiv (preprint; workshop paper, Regulatable Machine Learning at NeurIPS 2023)Towards Responsible Governance of Biological Design Tools2023-11-27 · publishedInspect the record ↗
- ArticulatesCenter for Security and Emerging Technology, Georgetown UniversityAI and Biorisk: An Explainer2023-12 · publishedInspect the record ↗
D5 Synthesis-order screening is a control point for AI-related biological risk
- Bears onarXiv (preprint, Emily H. Soice, Rafael Rocha, Kimberlee Cordova, Michael Specter, Kevin M. Esvelt)Can large language models democratize access to dual-use biotechnology?2023-06-06 · publishedInspect the record ↗
- Articulates · Earliest locatedarXiv (preprint, Jonas B. Sandbrink, Nuffield Department of Medicine, University of Oxford)Artificial intelligence and biological misuse: Differentiating risks of language models and biological design tools2023-06-24 · publishedInspect the record ↗
- Bears onarXiv (preprint; workshop paper, Regulatable Machine Learning at NeurIPS 2023)Towards Responsible Governance of Biological Design Tools2023-11-27 · publishedInspect the record ↗
D6 Public release of model weights increases biological misuse risk
- Articulates · Earliest locatedarXiv (preprint, Elizabeth Seger and 21 other authors; Centre for the Governance of AI and others)Open-Sourcing Highly Capable Foundation Models: An evaluation of risks, benefits, and alternative methods for pursuing open-source objectives2023-09-29 · publishedInspect the record ↗
- ArticulatesarXiv (preprint; workshop paper, Regulatable Machine Learning at NeurIPS 2023)Towards Responsible Governance of Biological Design Tools2023-11-27 · publishedInspect the record ↗
- Contests · Earliest locatedarXiv (preprint, Sayash Kapoor, Rishi Bommasani and others)On the Societal Impact of Open Foundation Models2024-02-27 · publishedInspect the record ↗
D7 Assessment must cover actors who already hold relevant skills
- Bears onarXiv (preprint, Jonas B. Sandbrink, Nuffield Department of Medicine, University of Oxford)Artificial intelligence and biological misuse: Differentiating risks of language models and biological design tools2023-06-24 · publishedInspect the record ↗
- Bears onCenter for Security and Emerging Technology, Georgetown UniversityAI and Biorisk: An Explainer2023-12 · publishedInspect the record ↗
- Articulates · Earliest locatedRAND Corporation (working paper WR-A3853-1), posted to arXivContemporary AI foundation models increase biological weapons risk2025-06-12 · publishedInspect the record ↗