The AI and Biological Weapons Threat
webCredibility Rating
High quality. Established institution or organization with editorial oversight and accountability.
Rating inherited from publication venue: RAND Corporation
A 2023 RAND empirical study directly relevant to catastrophic risk from AI misuse; provides early evidence on LLM dual-use risks in bioweapons contexts, informing debates about frontier model deployment safeguards and biosecurity policy.
Metadata
Summary
This RAND Corporation report examines the misuse risks of large language models (LLMs) in biological weapons development through a red-team methodology. Preliminary findings show that while LLMs haven't provided explicit weapon-creation instructions, they do offer guidance useful for planning biological attacks, including agent selection and acquisition strategies. The authors caution that AI's rapid advancement may outpace regulatory oversight, closing historical information gaps that previously hindered bioweapon development.
Key Points
- •LLMs did not generate explicit bioweapon instructions but provided actionable planning guidance including agent identification and distribution strategies.
- •In a plague pandemic scenario, an LLM assessed obtaining and distributing Yersinia pestis while estimating variables affecting projected death tolls.
- •In a botulinum toxin scenario, an LLM suggested aerosol delivery methods and proposed cover stories for acquiring dangerous biological agents.
- •AI advancement may outpace regulatory oversight, closing information gaps that previously caused biological attacks to fail.
- •The full real-world operational impact of LLMs on bioweapon planning remains an open research question requiring further study.
Cited by 4 pages
| Page | Type | Quality |
|---|---|---|
| AGI Development | -- | 52.0 |
| International AI Coordination Game Model | Analysis | 59.0 |
| AI Risk Interaction Matrix | Analysis | 65.0 |
| Bioweapons Risk | Risk | 91.0 |
Cached Content Preview
The Operational Risks of AI in Large-Scale Biological Attacks: A Red-Team Approach | RAND
Feb
MAR
Apr
01
2025
2026
2027
success
fail
About this capture
COLLECTED BY
Collection: Save Page Now Outlinks
TIMESTAMPS
The Wayback Machine - http://web.archive.org/web/20260301055948/https://www.rand.org/pubs/research_reports/RRA2977-1.html
Skip to page content
Toggle Menu
Site-wide navigation
Topics
Trending
International Trade
Iran
Artificial Intelligence
Chronic Diseases and Conditions
Emergency Preparedness
Topics
Children, Families, and Communities
Cyber and Data Sciences
Education and Literacy
Energy and Environment
Health, Health Care, and Aging
Homeland Security and Public Safety
Infrastructure and Transportation
International Affairs
Law and Business
National Security
Science and Technology
Social Equity
Workers and the Workplace
All Topics
Research & Commentary
Experts
About
Research Divisions
RAND's divisions conduct research on a uniquely broad front for clients around the globe.
U.S. research divisions
RAND Army Research Division
RAND Education, Employment, and Infrastructure
RAND Global and Emerging Risks
RAND Health
RAND Homeland Security Research Division
RAND National Security Research Division
RAND Project AIR FORCE
International research divisions
RAND Australia
RAND Europe
Services & Impact
Careers
Graduate School
Subscribe
Give
Cart
Toggle Search
Search termsSubmit
RAND
Research & Commentary
Research Reports
RR-A2977-1
In this report, the authors address the emerging issue of identifying and mitigating the risks posed by the misuse of artificial intelligence (AI)—specifically, large language models—in the context of biological attacks and present preliminary findings of their research. They find that while AI can generate concerning text, the operational impact is a subject for future research.
The Operational Risks of AI in Large-Scale Biological Attacks
A Red-Team Approach
Christopher A. Mouton, Caleb Lucas, Ella Guest
ResearchPublished Oct 16, 2023
Download PDF
Share on LinkedIn
Share on X
Share on Facebook
Email
The rapid advancement of artificial intelligence (AI) has far-reaching implications across multiple domains, including its potential to be applied in the development of advanced biological weapons. The speed at which AI technologies are evolving often surpasses the capacity of government regulatory oversight, leading
... (truncated, 8 KB total)73c1b835c41bcbdb | Stable ID: YmQyZDAyMz