Back to the MIT repository
4. Malicious Actors & Misuse2 - Post-deployment

Malicious use and abuse (biosecurity threats)

Many fear that generative AI could make the creation of biological weapons easier by providing access to critical knowledge and automated assistance to a wider range of actors to engage in malicious activities.

Source: MIT AI Risk Repositorymit731

ENTITY

1 - Human

INTENT

1 - Intentional

TIMING

2 - Post-deployment

Risk ID

mit731

Domain lineage

4. Malicious Actors & Misuse

223 mapped risks

4.2 > Cyberattacks, weapon development or use, and mass harm

Mitigation strategy

1. Implement rigorous, multi-layered technical and governance safeguards for advanced biological AI models (BDTs and LLMs), including mandatory pre-release safety evaluations, stringent data filtering to exclude high-risk sequences, and real-time monitoring of user requests to block the generation of hazardous biological constructs. 2. Develop a standardized, AI-enabled nucleic acid synthesis-screening system capable of detecting novel and augmented biological sequences generated by AI, thereby establishing dynamic chokepoint control that surpasses the limitations of current static, list-based screening protocols. 3. Bolster national and international biosecurity architecture by integrating AI into a comprehensive threat detection and response system, including continuous biosurveillance for anomalous bioengineering signatures and establishing secure, cross-sector intelligence-sharing platforms to monitor high-risk purchases and behavioral indicators of illicit biotechnological activity.