OpenAI warns of widening AI use gap between power users and the rest
A new report finds that advanced AI tools are being adopted unevenly across users and countries, raising concerns that differences in skills and access could deepen existing economic divides.
OpenAI has released a report highlighting growing disparities in how people and countries use advanced artificial intelligence tools, warning that uneven adoption risks widening economic and technological gaps.
According to the report, a relatively small group of “power users” is making far deeper use of advanced AI capabilities than the average user. OpenAI said the typical power user relies on around seven times more “advanced thinking” capabilities than a typical user. These capabilities include tools that go beyond basic question answering, such as data analysis, coding assistance and more autonomous, multi-step tasks.
For a lay reader, this distinction reflects how AI is used in practice. Many people use AI tools for simple queries, writing help or explanations. Power users, by contrast, are more likely to use features that analyse datasets, connect to external tools, write and debug code or automate complex workflows. These uses can save significant time and unlock new kinds of work, but they require familiarity, confidence and, in some cases, paid access.
The report said the gap is also visible at the country level. Looking across more than 70 countries with the highest ChatGPT usage, OpenAI found that some countries use three times more advanced thinking capabilities per person than others. This suggests that national differences in skills, infrastructure and policy are shaping how deeply AI is embedded in everyday work and learning.
In terms of overall scale, the United States and India lead in total number of users, reflecting their large populations and technology sectors. Smaller, high-income countries such as Singapore and the Netherlands rank highly for population penetration, meaning a larger share of their citizens use ChatGPT regularly.
The report also highlighted patterns that cut across income levels. Countries including Vietnam and Pakistan rank among the world’s top users of so-called agentic tools. These are features that allow AI systems to carry out more complex tasks, such as analysing data, working across connected applications or assisting with software development through tools like Codex. OpenAI said these countries show more than twice the per-person use of advanced tasks compared with the global average.
Agentic tools matter because they can change how work is organised. Instead of helping with isolated tasks, they can support end-to-end processes, from analysis to execution. Users who master these features may gain a disproportionate advantage in productivity and innovation.
OpenAI framed the findings as both a warning and a call to action. The company said it launched OpenAI for Countries last year to help governments move from basic AI use towards deeper, more strategic adoption. The initiative is intended to support public institutions in integrating AI into areas such as education, healthcare and public services.
This year, OpenAI said it plans to expand that work with new initiatives focused on education and health, AI skills training and certifications, disaster response and preparedness, cybersecurity and start-up accelerators. The aim is to lower barriers to advanced use and ensure that benefits are not confined to a narrow group of users or countries.
A central element of this effort is the Education for Countries programme. OpenAI said its first partners include Estonia, the United Arab Emirates, Greece, Jordan, Slovakia, Kazakhstan, Trinidad and Tobago, and Italy’s Conference of Rectors of Italian Universities, known as CRUI.
The programme is designed to combine several strands: expanded access to advanced AI tools, large-scale research into how AI affects education systems, training and certification for students and educators, and the creation of a global community of partner institutions. For education systems, this is intended to move beyond ad hoc experimentation towards coordinated, evidence-based adoption.
The report underscores a broader concern shared by policymakers and economists. AI has the potential to boost productivity and growth, but only if people have the skills and opportunities to use it effectively. If advanced capabilities remain concentrated among a small group of users and countries, existing inequalities could harden rather than shrink.
OpenAI did not argue that disparities are inevitable. Instead, it positioned them as a function of choices about access, training and institutional support. The data suggests that when users and organisations invest in learning how to use advanced tools, adoption can accelerate quickly, even in lower-income settings.
Related reading
- OpenAI pilots AI programme to support primary healthcare across Africa
- OpenAI says US data centre build is over halfway to 10GW target
- Google adds AI verification and ransomware safeguards to Workspace for Education
The challenge, the report implies, is scale. Teaching millions of people not just to use AI, but to use it well, requires sustained investment in education and capacity building. Without that effort, the gap between those who treat AI as a basic assistant and those who use it as a force multiplier is likely to grow.
As AI becomes more embedded in work, education and government, the question is no longer whether people use it, but how deeply. OpenAI’s warning is that depth, not headline adoption numbers, may be where the real divides emerge.
The Recap
- OpenAI published a report on global AI capability gaps.
- Typical power user uses about seven times more capabilities.
- OpenAI will expand programs this year across several sectors.
