Artificial intelligence is no longer a distant future; it is woven into the fabric of daily life, shaping how people find jobs, access healthcare, and participate in democracy. As these systems grow more powerful, a critical question has emerged: how do we ensure they are used for good and do not cause harm? The answer lies in "responsible AI," a concept that goes beyond writing rules on paper. It requires building actual institutions, creating transparent processes, and ensuring that diverse voices, especially those from civil society, have a seat at the table. Without a way to measure whether these safeguards exist and function in the real world, promises of ethical technology remain just words. To solve this, researchers have developed a new way to look at the global landscape, moving from abstract principles to concrete evidence of what nations are actually doing.

This is the story of the Global Index on Responsible AI, or GIRAI, specifically its 2026 edition. This report represents a major step forward in how the world evaluates the governance of artificial intelligence. Rather than simply ranking countries based on how much money they spend on technology or how many patents they hold, this index asks a different set of questions. It investigates whether governments have created laws that protect human rights, whether they are actively listening to the public, and whether they are avoiding the use of AI in ways that could damage society. The researchers behind the index gathered data from 135 countries and jurisdictions, creating a detailed map of where the world stands in its effort to govern this transformative technology responsibly.

The core of the index is built on a simple but powerful idea: responsible AI is not just about having a policy document; it is about what that document actually covers and how it is used. The researchers organized their findings into five key areas of life that AI touches: inclusion and diversity, ethics and sustainability, labor and skills, trust and safety, and the use of AI in public services. For each of these areas, they looked at three different angles. First, they examined the AI Policy pillar, which tracks the laws, strategies, and guidelines governments have written. Second, they measured the CSO Engagement pillar, which looks at how civil society organizations—like advocacy groups, unions, and community organizations—are involved in shaping these rules. Third, they assessed the Enabling Conditions pillar, which considers the broader environment, such as the strength of a country's democracy, the availability of digital tools for everyone, and the protection of personal data.

To build this picture, the team did not rely on self-reported surveys from governments alone. Instead, they deployed a global network of more than 160 researchers on the ground. These local experts scoured public records, legal databases, and news reports to find hard evidence. They looked for specific things: did a country pass a law that applies to all sectors, or just a few? Did the government hold public consultations before writing the rules? Did they create a plan with a budget to make the rules work? They also checked if civil society groups were actually active in these discussions. This method allowed them to distinguish between countries that merely talk about responsible AI and those that are building the machinery to make it happen.

A crucial and distinct part of this new edition is how it handles the worst-case scenarios. The researchers added a specific penalty for governments that deploy AI systems known to be unacceptable. These are technologies that pose a direct threat to human rights or democratic processes, such as social scoring systems, biometric surveillance used for repression, or AI-driven disinformation campaigns. If a country is found to be using such systems, its final score is reduced. This ensures that a country cannot receive a high rating for having good laws in one area while simultaneously using harmful technology in another. The penalty is proportional, meaning the more documented cases of misuse a country has, the more its score drops, reflecting the severity of the failure.

The 2026 edition also marks a significant evolution from the previous version of the index. The researchers refined their approach to better capture the reality of implementation. In the past, the focus was often on whether a framework existed. Now, the index digs deeper into the quality of that framework. It looks at whether the rules are enforceable, whether they include mechanisms for monitoring and evaluation, and whether they have been tested through public consultation. They also expanded the scope of what they count as "implementation." Instead of looking for just one government action, they now assess up to three different initiatives per topic, giving a fuller picture of how governments are trying to solve problems. This shift allows the index to see not just the existence of a policy, but the depth of a country's commitment to making it work.

The data collected reveals a complex global picture. The index covers 135 countries, representing a vast range of economic and political contexts. The researchers found that while many nations have started to draft policies, the gap between having a plan and putting it into practice remains wide. The inclusion of secondary data, such as statistics on internet access and the strength of the rule of law, helped them understand the context in which these policies operate. They discovered that in some places, the lack of basic digital infrastructure or the absence of strong civil society groups makes it difficult to implement responsible AI, regardless of how good the laws might look on paper.

One of the most important findings is that the index is not just a ranking tool; it is a diagnostic instrument. By breaking down the scores into specific dimensions and pillars, the report highlights exactly where a country is strong and where it is weak. A nation might have excellent laws regarding data privacy but fail completely in ensuring that workers are protected from AI-driven job displacement. Another might have a vibrant civil society but lack the government mechanisms to listen to them. This level of detail is designed to help policymakers, researchers, and advocates identify the specific gaps that need to be filled. It moves the conversation from "who is winning?" to "what needs to be fixed?"

The researchers were careful to ensure their results were robust. They subjected their methodology to a rigorous statistical audit to test whether small changes in how they calculated the scores would drastically alter the rankings. They found that the overall structure of the index holds up well under scrutiny. While the exact position of some countries in the middle of the rankings might shift slightly depending on the mathematical approach used, the leaders and the laggards remain consistent. This gives confidence that the index is measuring something real and stable, rather than just reflecting the quirks of a specific calculation method.

The report also acknowledges its own limitations. Measuring something as complex as responsible AI is inherently difficult. The index relies on publicly available evidence, which means that countries with less transparency or fewer resources to document their actions might appear to be doing less than they actually are. The researchers note that the field of AI governance is still young, and many countries are in the early stages of development. This is why the index places a heavy weight on government policy and a lighter weight on civil society engagement, recognizing that the primary driver of change right now is state action, even as the role of the public grows.

Ultimately, the Global Index on Responsible AI serves as a mirror for the world. It shows that while the principles of ethical technology are widely accepted, the work of turning those principles into reality is uneven. The 2026 edition provides a clearer, more detailed view of this landscape than ever before. It demonstrates that responsible AI is not a single destination but a continuous process of building frameworks, engaging communities, and avoiding harm. By making these efforts visible and comparable, the index offers a path forward for nations to learn from one another and to strengthen the governance of the technologies that will shape the future. The data is there, the methods are sound, and the message is clear: the time to move from aspiration to action is now.