Subject: Emerging Technologies

Mentor’s Comment

The rapid rise of Artificial Intelligence (AI) has pushed it from a commercial technology to a strategic national security asset. The debate intensified after American AI company Anthropic urged the U.S. government to classify Chinese AI labs like DeepSeek, Moonshot AI, and MiniMax as national security threats. The controversy reflects a deeper policy dilemma: Should AI be treated like nuclear technology requiring strict controls, or like a dual-use digital technology that thrives on open innovation? The issue has implications for military decision-making, global technological competition, and governance of autonomous systems.

Is AI becoming a national security technology comparable to nuclear weapons?

1. Dual-Use Technology: AI functions as a general-purpose technology used for civilian innovation and military operations. Unlike nuclear weapons, AI also drives sectors such as healthcare, finance, and digital governance.
2. Military Integration: AI models assist in accelerating the military “kill chain”, supporting target identification, intelligence analysis, and operational decisions.
3. Technological Diffusion: AI research occurs across universities, private firms, and open-source communities, enabling rapid global diffusion.
4. Comparative Argument: Nuclear non-proliferation succeeds due to scarcity of fissile material, whereas AI relies on widely accessible resources like data and computing.

What is AI model distillation and why is it controversial?

1. Model Distillation: Distillation involves training smaller AI models using the outputs of larger frontier models to replicate capabilities at lower computational cost.
2. Industrial-Scale Claims: Anthropic alleges 16 million interactions with its Claude model through around 24,000 accounts, suggesting systematic distillation efforts.
3. Strategic Advantage: Distillation enables competitors to achieve frontier-level performance at a fraction of the cost of original research.
4. Intellectual Property Issues: Companies argue distillation violates terms of service and proprietary model safeguards.

Why are export controls and technological restrictions facing limitations?

1. Circumvention of Restrictions: Export controls on advanced chips and inputs often face workarounds through alternative supply chains or domestic development.
2. Human Capital Mobility: AI researchers frequently work across countries, making technological containment difficult.
3. Diffusion of Knowledge: AI research spreads through academic publications, open-source models, and global conferences.
4. Policy Ineffectiveness: Restrictions may fail to prevent competitors from achieving comparable performance, as illustrated by emerging Chinese AI models.

Do corporate guardrails effectively regulate military uses of AI?

1. Corporate Governance Limits: Private companies can modify or remove safeguards when responding to government contracts.
2. Defense Integration: AI firms increasingly compete for military and national security contracts, accelerating integration into defence systems.
   1. Example: Some firms accept permissive contracts allowing military use of AI models, illustrating the competitive pressure in defence technology markets.
3. Regulatory Gap: Corporate policies alone cannot substitute state-led governance frameworks for military AI use.

Why does AI governance require international cooperation?

1. Inevitable Military Adoption: Armed forces globally are integrating generative AI into surveillance, cyber warfare, and autonomous systems.
2. Need for Global Norms: Effective regulation requires plurilateral commitments among states rather than unilateral corporate decisions.
3. Human Control: Governance frameworks must ensure meaningful human oversight in lethal decision-making systems.
4. Restrictions on Mass Surveillance: Global norms should prohibit large-scale civilian surveillance enabled by AI systems.

Way Forward: Strengthening Global Governance of AI in National Security

1. Multilateral AI Governance Framework: Establishes global rules for responsible AI deployment through platforms like the United Nations and the UNESCO which already adopted the Recommendation on the Ethics of Artificial Intelligence (2021) promoting transparency, accountability, and human rights protection.
2. AI Safety and Risk Management Regimes: Strengthens international cooperation through initiatives like the Global Partnership on Artificial Intelligence (GPAI) and the OECD AI Principles, which promote responsible AI innovation, democratic values, and safeguards against misuse.
3. Regulation of Military AI Systems: Develops binding norms on autonomous weapons through negotiations under the United Nations Convention on Certain Conventional Weapons (CCW), focusing on meaningful human control over lethal autonomous weapons systems (LAWS).
4. Global Technology Export and Monitoring Mechanisms: Expands export-control regimes such as the Wassenaar Arrangement to include AI algorithms, advanced chips, and surveillance systems to prevent uncontrolled proliferation.
5. Data Governance and Digital Rights Protection: Aligns AI regulation with frameworks such as the European Union AI Act, which classifies AI systems by risk level and restricts high-risk surveillance technologies.
6. International Research Collaboration: Promotes open but secure collaboration among states, universities, and companies through forums like the G20 and World Economic Forum, ensuring innovation while maintaining safeguards.
7. India’s Strategic Role: India can leverage platforms such as the BRICS, Quad, and G20 to push for ethical AI standards, responsible military use, and inclusive technological governance.

Conclusion

Artificial Intelligence is transforming the intersection of technology, geopolitics, and national security. Unlike nuclear technology, AI cannot be easily contained due to its open research ecosystem, global talent mobility, and digital diffusion. Effective governance therefore requires international norms, state-led oversight, and responsible corporate practices to balance innovation with security.