Bioterrorism
AIs with knowledge of bioengineering could facilitate the creation of novel bioweapons and lower barriers to obtaining such agents.
ENTITY
2 - AI
INTENT
1 - Intentional
TIMING
2 - Post-deployment
Risk ID
mit341
Domain lineage
4. Malicious Actors & Misuse
4.2 > Cyberattacks, weapon development or use, and mass harm
Mitigation strategy
1. Implement multi-layered technical and governance safeguards on advanced biological models, including mandatory user verification protocols and access controls to deter misuse. This must be coupled with rigorous technical defenses such as data filtering and model unlearning to remove high-consequence biological design knowledge while preserving beneficial capabilities. 2. Mandate and develop a standardized, AI-enabled nucleic acid synthesis-screening system across all providers. This system must integrate sophisticated AI algorithms to detect novel or augmented sequences that may not appear on existing restricted lists, and must be designed to circumvent obfuscation techniques such as split ordering of dangerous genetic fragments. 3. Prioritize the development and integration of defensive AI capabilities to enhance global biosecurity. This includes accelerating AI-driven biosurveillance for the rapid detection of engineered outbreaks and leveraging AI for the swift design and optimization of medical countermeasures and treatments against novel bioweapons.
ADDITIONAL EVIDENCE
Engineered pandemics from AI-assisted bioweapons pose a unique challenge, as attackers have an advantage over defenders and could constitute an existential threat to humanity.