RAND представил стратегию многоуровневой защиты от биологического оружия с использованием ИИ

RAND Outlines Layered Defense Strategy to Mitigate AI-Enabled Bioweapon Risk For Release Tuesday August 18, 2026 No single safeguard can prevent an actor from using artificial intelligence to help build a biological weapon, but a new RAND report identifies nine interventions that, layered together, could meaningfully lower the risk of a high-consequence AI-enabled biological attack. The report, “Building a Defense-in-Depth Biosecurity Strategy for the AI Era,” maps the full range of actions that could lead to an AI-enabled biological attack, and examines where and how different safeguards can most effectively intervene to prevent misuse at each stage. “AI is lowering the technical, operational and motivational barriers that have kept bioweapons attacks out of reach for most actors. Unfortunately, today's safeguards are fragmented across companies, governments and countries and are not designed to work together,” said Steph Guerra, head of AI x Bio at RAND and lead author of the report. “This strategy shows how layered, mutually reinforcing safeguards can close those gaps and adapt as AI capabilities evolve.” Different actors require different defenses, according to the report. A lone individual with limited resources or technical expertise may be disrupted by cutting off access to key materials and information. A well-funded group or state program is a harder target and may be more likely to be discouraged by raising the costs of an attack—or the odds of getting caught. Detection, not just restriction, is central to the report's approach. Because determined actors with enough resources can work around barriers, the report argues that an effective strategy also requires detection systems that can flag suspicious patterns that look harmless alone but form a clear warning sign when viewed together. Those capabilities create opportunities for both disruption and deterrence, but require a shared infrastructure that no single company, agency, or nation can build alone. The nine mitigations in this defense-in-depth strategy work by integrating disruption, detection, and deterrence approaches to identify and counteract actors who might otherwise bypass traditional controls. Three restrict access to dangerous information and AI tools, including safeguards for open-source models and managed access programs and platforms for sensitive AI systems. Another focuses on expanding product and customer screening for biological precursors that can be used to make biological weapons. Three mitigations aim to deter attacks through early warning, attribution and rapid outbreak response. And the final two focus on detecting misuse through real-time monitoring of AI model use and sharing warning signs across companies, agencies and governments. The report cautions that some safeguards will take years to build, test and implement, arguing that waiting until AI's risks are undeniable would leave a dangerous gap between what the technology can do and the protections in place to prevent misuse. The report also finds that several mitigations lose their effectiveness if the U.S. acts alone, making international coordination critical alongside collaboration between government and industry. “We don't know exactly how fast AI capabilities in biology will advance, or which actor will try to exploit them first, but we do know that the window for building preventive infrastructure is open now,” said Sella Nevo, director of RAND's Center on AI, Security, and Technology. “This strategy is designed to be robust across a range of future possibilities." The research was conducted by RAND's Center on AI, Security, and Technology, part of RAND Global and Emerging Risks. Other authors include Aurelia Attal-Juncqua, John P. Tarangelo, Casey Aveggio, Katie Dammer and David Glickstein.