Former White House homeland security adviser says AI lets more actors build bioweapons and urges deterrence by resilience
Foreign Affairs · Essay (paywalled)
Read the source (opens in a new tab)- What happened
- Elizabeth Sherwood-Randall, White House homeland security adviser from 2021 to 2025, published an essay in Foreign Affairs on August 18, 2026 on AI-enabled biological threats.
- Key points
- She writes that "The United States is unprepared for these threats," and that deterrence by threatened retaliation fails once AI lets rogue states, terrorist groups and individuals develop bioweapons.
- She cites April 2025 research in which language models "consistently outperformed Ph.D.-level virologists" on lab-knowledge tests, and says cloud labs and contract manufacturers face no regulation.
- She calls for a strategy that "accepts that biological attacks are likely": nationwide early warning, faster attribution, biotech as critical infrastructure and a national readiness reserve.
- She urges mandatory DNA synthesis screening, global biosecurity summits, and tiered access to the most capable biological AI models, saying AI companies' bio safeguards "remain voluntary."
- Evidence
- A signed policy essay by a former senior US official; it is an argument. Its attack scenarios are hypothetical, and it reports no new data.