Destabilising political impacts from AI systems

Mitigation strategy

1. Strengthen Information Ecosystem and Cognitive Resilience Implement a comprehensive, multi-layered approach to counteract AI-driven disinformation, which includes measures to curtail the generation of malicious content, constrain its dissemination on digital platforms, and foster extensive digital literacy and cognitive resilience campaigns to empower society against malicious influence. 2. Mandate Transparency and Human Oversight in Electoral AI Establish stringent regulatory and operational requirements for the use of AI in electoral processes, necessitating continuous testing and evaluation, mandating the disclosure of training datasets by vendors to preempt bias, and ensuring human-in-the-loop decision-making for all critical election-related content and actions. 3. Promote International Confidence-Building Measures for Military AI Establish a multilateral framework for Confidence-Building Measures (CBMs) in military AI systems, such as increasing transparency regarding Testing and Evaluation (T\&E) processes, and facilitating international information-sharing to reduce the risk of miscalculation and prevent unintended escalation between states.