Caleb WithersResearch Associate, Technology and National Security Program (CNAS)
Think tank · Policy · Pennsylvania Ave NW, Washington · Est. 2007
CNAS AI Security & Stability is an Artificial Intelligence-focused, cross-program research effort within the Center for a New American Security (CNAS). The project’s stated purpose is to inform government decision-making on AI policy issues that affect the future of AI development, with an emphasis on national-security-relevant advanced AI capabilities, compute governance, and how U.S. military processes can ensure safe, secure, and trusted AI.
The project also includes analysis of U.S.-China AI strategic stability dynamics and the implications of U.S. economic-security tools for AI diffusion and proliferation. CNAS frames the initiative as a cross-program effort drawing on its Technology and National Security, Defense, Indo-Pacific Security, and Energy, Economics, and Security programs, with public reports and policy briefs for policymakers. The initiative is supported (per CNAS) by Coefficient Giving, and CNAS previously announced a multi-year AI safety and stability project (2012-era naming differs) supported by Open Philanthropy.
No indexed openings right now. Check the careers page ↗.
This essay series, The New American Defense Industrial Base, featured expert practitioners with experience in government, industry, and finance.
Sharper: China's AI RiseDeveloping strong, pragmatic and principled national security and defense policies.
Indo-Pacific Stability Depends on Eastern Africa’s ProsperityDeveloping strong, pragmatic and principled national security and defense policies.
Manufacturing a New National SecurityDeveloping strong, pragmatic and principled national security and defense policies.
Beneath the Surface: Overlooked Cyber Security ThreatsDeveloping strong, pragmatic and principled national security and defense policies.
We Cannot Let the Lebanon-Israel Breakthrough Slip AwayDeveloping strong, pragmatic and principled national security and defense policies.
Why Empowering Private Companies to Fight Cybercrime Had to HappenDeveloping strong, pragmatic and principled national security and defense policies.
The Impossible Middle EastDeveloping strong, pragmatic and principled national security and defense policies.
From Ukraine to Iran, Attacks on Civilian Sites Put Us All at RiskDeveloping strong, pragmatic and principled national security and defense policies.
Shutdown Reforms Constitutional Blind Spot The Two Year ClauseBeijing Sees One Theater Washington Should TooThe Risks Of Downplaying Chinas MilitaryA CNAS Insights piece discussing a reported incident involving frontier AI models escaping isolated testing environments and what it implies for control and safety risks.
Red LinesCNAS report analyzing national security risks of China’s advanced AI systems.
Noteworthy | Promoting Advanced Artificial Intelligence Innovation and SecurityA CNAS Noteworthy edition commenting on an executive order on AI oversight released in June 2026, including discussion of vetting advanced AI models for national security risks.
CNAS Insights | Trump Should Talk to Xi About Military AICNAS Insights piece arguing for attention to military AI as an area where U.S.-China could agree on measures to manage accidents and enhance stability without unrealistic constraints on innovation.
Off TargetWorking paper on AI alignment challenges for national security, discussing the DoD’s stated AI acceleration direction and implications for strategic risk.
CNAS Expands AI Policy Research with New AppointmentsCNAS press release announcing three staff hires as it expands its multiyear AI Security and Stability initiative (including Daniel Remler, James Sanders, and Liam Epstein).