AI Security Preparedness
Recent Developments
- Anthropic’s Claude Mythos demonstrated advanced cybersecurity capabilities beyond human experts in certain tasks. The growing access to Mythos-class AI has raised concerns over security of critical infrastructure.
Need for AI
- Critical Infrastructure Security: AI can identify and prevent threats targeting financial, power, and governance systems.
- Detection of Hidden Vulnerabilities: AI can discover previously unknown software flaws missed by human experts.
- Faster Cyber Defence: AI enables security responses at machine speed against rapidly evolving attacks.
- Protection of Digital Public Infrastructure: AI strengthens resilience of large-scale digital systems and services.
- Managing Emerging Cyber Threats: AI helps anticipate and counter increasingly sophisticated cyber risks.
India’s Preparedness Gap
- Legacy Technology Dependence: Many critical systems continue operating on outdated and fragmented back-end infrastructure.
- Absence of AI Safety Institution: India lacks a dedicated body to assess frontier AI risks.
- Limited Testing Capacity: Advanced AI models are not adequately evaluated against Indian vulnerabilities.
- Cybersecurity Workforce Shortage: A large skills gap weakens the country’s cyber defence capabilities.
- Slow Vulnerability Response: Patch cycles remain slower than the speed of AI-enabled attacks.
- Regulatory Vacuum: No dedicated framework exists for oversight of high-risk frontier AI models.
Way Forward
- Establish AI Safety Institute: Create a dedicated institution to evaluate frontier AI risks.
- Strengthen International Cooperation: Partner with trusted countries for access to defensive AI capabilities.
- Develop AI Accountability Framework: Mandate disclosure of risks and capabilities of advanced AI systems.
- Modernise Critical Infrastructure: Upgrade legacy systems in banks and government institutions.
- Invest in Cybersecurity Capacity: Expand skilled workforce and strengthen institutional preparedness.
- Develop Sovereign Defensive AI: Build indigenous AI tools for threat detection and rapid response.
- Lead Global AI Governance: Promote international norms for responsible release of powerful AI models.
- Ensure Whole-of-Government Coordination: Drive AI security preparedness through coordinated national-level action.
