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

Democratizing access to dual-use technologies

Access to dual-use technologies can become easier because of GPAI model pro- liferation (in particular, open-source or open-weights models). Non-experts can use such dual-use-capable systems at a minimal cost [194, 100]. Improved model capabilities also contribute to dual-use risks posed by malicious actors. For example, an open-source base model for generating high quality sequence data can be modified to generate candidate protein sequences for toxin synthesis [29].

Source: MIT AI Risk Repositorymit1167

ENTITY

1 - Human

INTENT

3 - Other

TIMING

2 - Post-deployment

Risk ID

mit1167

Domain lineage

4. Malicious Actors & Misuse

223 mapped risks

4.0 > Malicious use

Mitigation strategy

1. Mandate comprehensive, capability-focused Model Evaluations and Red-Teaming prior to and following the release of dual-use GPAI models, specifically targeting chemical and biological misuse pathways to assess the marginal risk introduced by the model's capabilities (e.g., according to NIST AI 800-1 guidelines). 2. Implement a Standardized, AI-Enabled Screening System for Nucleic Acid Synthesis to serve as a critical choke point defense, capable of detecting novel or AI-generated toxicological and pathogenic sequences that circumvent current list-based screening protocols. 3. Establish and enforce an industry-wide Model Artifact Trust and Traceability Framework for open-weight foundation models to ensure cryptographic signing, auditability of model components and weights, and clear, standardized technical documentation to facilitate post-deployment monitoring and attribution of misuse.