The arrival of ChatGPT in November 2022 sparked a global conversation about the potential and risks of artificial intelligence (AI). This excitement led to new research and policy initiatives aimed at addressing large-scale AI risks. However, these efforts have also created a complex debate about how to define and prioritize AI hazards.
In light of this, the Center for a New American Security (CNAS) released Catalyzing Crisis: A Primer on Artificial Intelligence, Catastrophes, and National Security. This report aims to help policymakers understand the critical importance of mitigating AI’s catastrophic risks for national security. It clarifies what these risks are, distinguishes them from other threats, and introduces key dimensions of AI safety that will shape future risks.
In the report, the authors suggest that powerful technologies like AI can pose significant dangers. AI has wide-ranging applications, including in biosecurity and military systems, making proactive risk management essential. Historical incidents in finance, biosecurity, cybersecurity, and nuclear control suggest potential future AI-related catastrophes. These could range from AI-accelerated biological weapons to financial meltdowns or accidental weapon exchanges.
Despite concerns, deep learning-based AI tools are still in their early stages in high-risk domains. This gives the United States a chance to develop AI responsibly. The report outlines four dimensions of AI safety that shape catastrophic risks and suggests that AI catastrophes are more likely to originate from China due to its rapid AI development and crisis management practices.
To address these risks, the report recommends precise terminology, thorough exploration of AI integration into high-impact domains, enhanced testing, and planning for AI-related catastrophes abroad. By managing AI’s catastrophic risks, the U.S. can advance AI’s benefits for security and competitiveness.
For more information, check out some HSDL documents related to artificial intelligence and national security, as well as its In Focus topic on Artificial Intelligence.
