Georgia is taking an unconventional approach to AI governance: instead of rushing to pass comprehensive laws, the country is building a four-phase roadmap that starts with defining a national vision, then moves through strategy, practical guidelines, and finally institutional architecture. This measured approach, based on UNESCO's AI Readiness Assessment Methodology (RAM), offers a potential model for smaller nations and developing economies trying to regulate AI without stifling innovation.

Why Is Georgia Starting With Vision Instead of Legislation?

Georgia currently lacks a dedicated national AI strategy, AI-specific legislation, or formal institutional framework for AI governance. Rather than viewing this as a disadvantage, experts argue it presents an opportunity to design an AI ecosystem deliberately, aligned with international standards and national needs.

The assessment found that responsibilities related to digital transformation, innovation, data protection, and cybersecurity remain scattered across different government institutions. This fragmentation is common in countries early in their AI governance journey, but it also means Georgia can learn from international experience before locking in permanent structures.

International precedent supports this phased approach. The European Union, before adopting its landmark AI Act, first developed broader policy vision through initiatives like the 2018 Coordinated Plan on Artificial Intelligence and the 2020 White Paper on Artificial Intelligence. These documents served as consultation frameworks to discuss governance options before binding regulation was introduced.

What Are Georgia's Four Phases of AI Governance?

• Phase 1 - National Vision: Georgia would develop a National AI White Paper to translate readiness assessment findings into a national conversation. This document would answer fundamental questions about what AI future Georgia wants to build, whether the focus should be on responsible adoption, AI-enabled public services, sector-specific capabilities, or a competitive regional ecosystem.
• Phase 2 - Strategy and Standards: A National AI Strategy would transform priorities into concrete objectives, timelines, and measurable actions. In parallel, Georgia could develop non-binding national standards and guidelines on responsible AI use, aligned with international frameworks like UNESCO's Recommendation on the Ethics of Artificial Intelligence.
• Phase 3 - Practical Experience: Regulatory experimentation mechanisms such as AI sandboxes and controlled pilots would allow policymakers to test approaches in real-world settings, identify emerging risks, and determine what institutional or regulatory measures may be needed in the future.
• Phase 4 - Institutional Architecture: After gaining practical experience, Georgia would establish sustainable institutional structures and capacity for long-term AI governance, designed around governance needs identified through earlier phases rather than created in isolation.

This phased approach reflects lessons from countries at various stages of AI governance maturity. Israel has advanced a flexible, principles-based approach centered on ethical guidelines and voluntary standards, explicitly reserving the option of future legislation if cross-sectoral risks emerge. Japan relies on non-binding AI Guidelines for Business that articulate human-centric, ethical, and governance principles such as safety, fairness, transparency, and accountability, while allowing space for gradual evolution toward binding rules where justified. Singapore developed nationally endorsed frameworks like the Model AI Governance Framework and practical tools like AI Verify, which provide shared definitions and operational guidance without imposing legal obligations.

How Can Non-Binding Standards Protect Against AI Harms?

For Georgia, non-binding national standards and guidelines could promote responsible AI use by focusing on specific areas of concern. These would include transparency and explainability, accountability, human oversight, risk assessment, data governance, and fairness.

This approach is particularly relevant given the documented harms of unchecked AI deployment. In India, for example, 47% of adults have either been victims of or know someone targeted by AI voice-cloning or deepfake scams, nearly double the global average of 25%. Among Indian victims, 83% suffered direct monetary losses, with almost half losing more than 50,000 rupees.

Beyond financial fraud, unchecked AI poses broader societal risks. AI-driven hostile content can reach targeted audiences for as low as $0.07 per view, making the weaponization of synthetic misinformation a highly scalable threat to pluralistic societies. Automated systems are increasingly deciding life-altering outcomes such as who gets a job, a loan, a medical bed, or an education, often without mandatory human oversight or accountability.

What Institutional Model Should Georgia Adopt?

International experience demonstrates that there is no single institutional model for AI governance. Some countries establish dedicated AI bodies, while others rely on coordination mechanisms across existing institutions. The appropriate model depends on national priorities, administrative structures, regulatory objectives, and the maturity of the AI ecosystem.

For Georgia, the objective should not simply be creating new institutions, but ensuring that institutions have the mandate, expertise, and coordination mechanisms required to govern AI effectively. Institutions should be designed around governance needs identified through practice, rather than created before those needs are fully understood.

Clear institutional mandates will be essential. Georgia should define responsibilities for key governance functions, including who leads national AI policy and who monitors emerging risks. This coordination is particularly important given the velocity gap between AI innovation and traditional legislative processes. AI evolves at the speed of startup culture and mathematical innovation, while democratic lawmaking moves slowly. By the time landmark laws such as the European Union's Artificial Intelligence Act or the United Kingdom's Online Safety Act are passed, the technical harms they were built to combat have often already mutated.

How Does Georgia's Approach Compare to India's Regulatory Framework?

While Georgia is still in the readiness phase, India has already implemented a robust, multi-sectoral legislative and policy architecture. India's Information Technology Amendment Rules 2026 directly target synthetic content, defining Synthetically Generated Information (SGI) as computationally created or altered audio, visual, or audio-visual content appearing authentic. The rules mandate a three-hour takedown requirement for internet intermediaries and social media platforms to remove unlawful SGI upon notice, shrinking to two hours for non-consensual deepfake or sexually explicit content.

India's framework also includes the IndiaAI Mission and India AI Governance Guidelines, which mandate that AI systems adhere to seven principles: Trust, People First, Innovation over Restraint, Fairness and Equity, Accountability, Understandable by Design, and Safety, Resilience and Sustainability. The guidelines establish specialized national bodies including the AI Governance Group, the Technology and Policy Expert Committee, and an AI Safety Institute.

Additionally, India's Digital Personal Data Protection Act 2023 restricts data scraping or usage of personal data to train Large Language Models (LLMs) or neural networks without explicit, unambiguous consent, with strict financial penalties for non-compliance. The country's Draft Regulations for Use of Artificial Intelligence in Courts 2026 permits administrative automation but strictly prohibits AI from exercising core adjudicatory functions like drawing judgments, explicitly holding judicial officers responsible for errors caused by "black box" algorithms or AI "hallucinations".

Georgia's phased approach offers a different model, one that prioritizes building consensus and institutional capacity before imposing binding legal requirements. This strategy may prove particularly valuable for smaller economies with limited regulatory infrastructure, allowing them to learn from both successes and failures in more mature AI governance systems before committing to permanent legal structures.