Analysis

All Analysis

Filter by

1 - 8 of 8 results

Benchmarking as a Path to International AI Governance

AI benchmarking could serve as a practical foundation for international AI governance related to security issues, enabling confidence-building measures between states while advancing U.S. interests in global AI standardization

Commentary by Ian Reynolds — August 5, 2025

Toward Reliable AI, from the Bottom Up

The Trump administration’s AI Action Plan calls for reliable AI, but achieving this requires bottom-up-driven benchmarking that engages civil society, not just tech companies and government, to ensure AI performs safely in critical applications such as national security.

Commentary by Ian Reynolds — July 28, 2025

Asking AI How to Respond to a Foreign Policy Decision

AI models have built-in biases that factor into their decisionmaking. These biases should be mitigated before integrating them into foreign policy processes.

Analysis by Yasir Atalan, Benjamin Jensen, and Ian Reynolds — April 17, 2025

Hawkish AI? Uncovering DeepSeek’s Foreign Policy Biases

DeepSeek’s foreign policy recommendations reveal a troubling trend toward escalation in crisis scenarios—raising concerns about AI-induced bias and the risks of integrating off-the-shelf models into national security workflows.

Commentary by Ian Reynolds, Benjamin Jensen, and Yasir Atalan — April 16, 2025

AI Biases in Critical Foreign Policy Decisions

Can AI be trusted for high-stakes foreign policy decisions? Our new study uncovers how biases shape AI’s approach to conflict escalation and other critical foreign policy decision domains.

Commentary by Yasir Atalan, Ian Reynolds, and Benjamin Jensen — February 26, 2025

Calibrating NATO’s Vision of AI-Enabled Decision Support

In the future, NATO will be the global test case for the integration of artificial intelligence with military operations, setting the standards for interoperability, organizational applications, and responsible use of AI. 

Commentary by Ian Reynolds and Yasir Atalan — July 8, 2024