Artificial intelligence and biological misuse: Differentiating risks of language models and biological design tools
arXiv v1 (24 June 2023). The arXiv abstract page for 2306.13952v1 records the v1 submission as 24 June 2023 and a latest version v8 of 23 December 2023 (`arxiv-2306-13952-abs`). Only v1 was retrieved and quoted.
- Issued by
- arXiv (from the version record)
- Authors credited
- No authorship claim recorded yet. The source records list arXiv (Cornell University) and arXiv (preprint, Jonas B. Sandbrink, Nuffield Department of Medicine, University of Oxford).
- Dates
- Published 2023-06-24
- Revised 2023-12-23
The arXiv abstract page for 2306.13952v1 records the v1 submission as 24 June 2023 and a latest version v8 of 23 December 2023 (`arxiv-2306-13952-abs`). Only v1 was retrieved and quoted.
- Exact source
- arXiv abstract page for 2306.13952v1 (submission history) arXiv (Cornell University) · publication date unknown · Retrieved for the candidate pack on 2026-10-28 (SHA-256 f05b805a071dd4ad81ec84279296ed530c8a4b679f8d05ffa4760d24a79d8440). Abstract page as served on 2026-10-28; used only for the v1 submission date and the version history Retrieved again at admission on 2026-10-30 (SHA-256 f05b805a071dd4ad81ec84279296ed530c8a4b679f8d05ffa4760d24a79d8440, the same bytes). Its wording can change; earlier wording is not authenticated.
- Artificial intelligence and biological misuse: Differentiating risks of language models and biological design tools arXiv (preprint, Jonas B. Sandbrink, Nuffield Department of Medicine, University of Oxford) · publication date unknown · Retrieved for the candidate pack on 2026-10-28 (SHA-256 1ab85d8647f2d23079d35eb8162b70b3f970094106808ef4291fb89615db0afa). arXiv v1 of 24 June 2023 (the article was revised to v8 on 23 December 2023; this record cites v1 only) Retrieved again at admission on 2026-10-30 (SHA-256 1ab85d8647f2d23079d35eb8162b70b3f970094106808ef4291fb89615db0afa, the same bytes).
- Earliest located in Arguments about AI and biological risk for D4 Biological design tools pose risks distinct from those of language models →
- Earliest located in Arguments about AI and biological risk for D5 Synthesis-order screening is a control point for AI-related biological risk →
Appears in
- Arguments about AI and biological risk →4 records on this document
Versions
sandbrink-ai-biological-misuse: 1 recorded version, oldest first, published by arXiv. A version is shown as replacing another only where a record says so. Date order alone never implies it.
- 2023-12-23 · revisedarXiv v1 (24 June 2023) (this version)
This version
Earliest version recorded here; no replacement is recorded.
Documentary relations
Each relation is listed separately. Citing, crediting and reporting the use of a method are different acts, and none of them shows agreement or influence by itself.
Articulates · 3
Propositions whose criterion a passage of this document meets. The record says what the text states, in its register; it is not evidence that the proposition is true.
- Currently available AI systems increase an actor's biological-misuse capabilitySandbrink's preprint on AI and biological misuse (arXiv v1, 24 June 2023) states that large language models such as GPT-4 are already able to provide dual-use information that could have enabled historical biological weapons efforts to succeed.2023-06-24 · publication · Stated in source · Supported Inspect claim & evidence ↗Currently available AI systems increase an actor's biological-misuse capability: propositions page →
- Biological design tools pose risks distinct from those of language modelsSandbrink's preprint (arXiv v1, 24 June 2023) differentiates large language models from biological design tools, stating that language models in particular lower barriers to biological misuse while design tools expand the capabilities of sophisticated actors.2023-06-24 · publication · Stated in source · Supported Inspect claim & evidence ↗Biological design tools pose risks distinct from those of language models: propositions page →
- Synthesis-order screening is a control point for AI-related biological riskSandbrink's preprint (arXiv v1, 24 June 2023) states that enhanced screening of gene synthesis is among the measures key to reducing the risks it describes.2023-06-24 · publication · Stated in source · Supported Inspect claim & evidence ↗Synthesis-order screening is a control point for AI-related biological risk: propositions page →
Bears on · 1
Propositions a passage comes close to without meeting the criterion; each record names what is missing.
- Assessment must cover actors who already hold relevant skillsSandbrink's preprint (arXiv v1, 24 June 2023) states that biological design tools will expand the capabilities of sophisticated actors, and proposes interventions to deter biological misuse by such actors.2023-06-24 · publication · Stated in source · Supported Inspect claim & evidence ↗Assessment must cover actors who already hold relevant skills: propositions page →
Not recorded for this document: cites · cited by · explicitly credits · reports using a method from · method reported as used by · applies parts of · parts applied by · discusses · reports evaluation of. A missing record does not show that the relation does not exist.