{"entities":[{"id":"soice-llm-dual-use-2023-06","name":"Can large language models democratize access to dual-use biotechnology?","subtype":"Preprint","description":"arXiv v1 (6 June 2023). The arXiv abstract page for 2306.03809v1 records the v1 submission as 6 June 2023 (`arxiv-2306-03809-abs`). The PDF text layer carries no date.","sourceIds":["source-ab-soice-2023","source-ab-arxiv-2306-03809-abs"]},{"id":"ab-prop-current-system-uplift","name":"Currently available AI systems increase an actor's biological-misuse capability"},{"id":"ab-prop-physical-bottleneck","name":"The binding barrier is physical execution rather than information"},{"id":"ab-prop-synthesis-screening-control-point","name":"Synthesis-order screening is a control point for AI-related biological risk"},{"id":"arxiv","name":"arXiv"}],"claims":[{"id":"ab-arg-d1-soice-articulates","statement":"The preprint “Can large language models democratize access to dual-use biotechnology?” (arXiv v1, 6 June 2023) states that its results suggest that large language models will make pandemic-class agents widely accessible as soon as they are credibly identified, even to people with little or no laboratory training.","derivation":"explicit","resolution":"supported","editorialState":"seed","evidenceIds":["evidence-ab-arg-d1-soice-articulates-1","evidence-ab-arg-d1-soice-articulates-2","evidence-ab-arg-d1-soice-articulates-3"],"attributes":{"thought":{"roles":["earliest-located"],"register":"nonfiction","qualifier":"States the claim as a suggestion drawn from a one-hour classroom exercise."}}},{"id":"ab-arg-d2-soice-contests","statement":"The preprint \"Can large language models democratize access to dual-use biotechnology?\" (arXiv v1, 6 June 2023) states that the agents it discusses would become widely accessible as soon as they are credibly identified, even to people with little or no laboratory training, against the premise that laboratory execution is the binding barrier.","derivation":"explicit","resolution":"supported","editorialState":"seed","evidenceIds":["evidence-ab-arg-d2-soice-contests-1","evidence-ab-arg-d2-soice-contests-2"],"attributes":{"thought":{"roles":["contesting","earliest-located"],"register":"nonfiction","qualifier":"The contesting point is one clause of a conclusion drawn from a one-hour classroom exercise.","engagement":{"kind":"engages-premise","rationale":"The sentence addresses the premise about laboratory training directly rather than asserting only a contrary conclusion."}}}},{"id":"ab-arg-d5-soice-context","statement":"The preprint of 6 June 2023 states that promising nonproliferation measures include pre-release evaluations of language models by third parties, curation of training datasets, and verifiable screening of all DNA generated by synthesis providers.","derivation":"explicit","resolution":"supported","editorialState":"seed","evidenceIds":["evidence-ab-arg-d5-soice-context-1"],"attributes":{"thought":{"roles":["context"],"register":"nonfiction","qualifier":"Screening is one of three measures the preprint proposes, not a sole control point.","missing":"The passage proposes screening among three promising measures; it does not state that mitigation of AI-related biological risk depends on screening, which is the proposition’s criterion."}}}],"sources":[{"id":"source-ab-soice-2023","title":"Can large language models democratize access to dual-use biotechnology?","authors":["arXiv (preprint, Emily H. Soice, Rafael Rocha, Kimberlee Cordova, Michael Specter, Kevin M. Esvelt)"],"publisher":"arXiv (preprint, Emily H. Soice, Rafael Rocha, Kimberlee Cordova, Michael Specter, Kevin M. Esvelt)","publicationDate":null,"version":"Retrieved for the candidate pack on 2026-10-28 (SHA-256 78647f96bb8fed0e988de8807caa503cebd4b1148f47ad9d518b9cc401a2bbfe). arXiv v1 of 6 June 2023 (PDF at /pdf/2306.03809v1) Retrieved again at admission on 2026-10-30 (SHA-256 78647f96bb8fed0e988de8807caa503cebd4b1148f47ad9d518b9cc401a2bbfe, the same bytes).","evidenceUrl":"https://arxiv.org/pdf/2306.03809v1","language":"en"},{"id":"source-ab-arxiv-2306-03809-abs","title":"arXiv abstract page for 2306.03809v1 (submission history)","authors":["arXiv (Cornell University)"],"publisher":"arXiv (Cornell University)","publicationDate":null,"version":"Retrieved for the candidate pack on 2026-10-28 (SHA-256 28e8f611fd8130bd519b35594c214b1fbf23c4fee9d528708cd24e7cc2ddf691). Abstract page as served on 2026-10-28; used only for the v1 submission date Retrieved again at admission on 2026-10-30 (SHA-256 90fb4537c1ebbf342ac51f06924097f1bb97b9ae15271593cd85744c567f5cc0). Its wording can change; earlier wording is not authenticated.","evidenceUrl":"https://arxiv.org/abs/2306.03809v1","language":"en"}],"evidence":[{"id":"evidence-ab-arg-d1-soice-articulates-1","sourceRole":"primary","support":"supports","verification":"machine_checked","retrievalRoute":"direct"},{"id":"evidence-ab-arg-d1-soice-articulates-2","support":"supports","verification":"machine_checked","retrievalRoute":"direct"},{"id":"evidence-ab-arg-d1-soice-articulates-3","support":"supports","verification":"machine_checked","retrievalRoute":"direct"},{"id":"evidence-ab-arg-d2-soice-contests-1","sourceRole":"primary","support":"supports","verification":"machine_checked","retrievalRoute":"direct"},{"id":"evidence-ab-arg-d2-soice-contests-2","support":"supports","verification":"machine_checked","retrievalRoute":"direct"},{"id":"evidence-ab-arg-d5-soice-context-1","sourceRole":"primary","support":"supports","verification":"machine_checked","retrievalRoute":"direct"}],"at":{"entities":[1250,1254,1255,1258,1261],"claims":[3412,3413,3414],"sources":[1063,1064],"evidence":[4698,4699,4700,4701,4702,4703]},"version":"1lg47t7"}