A new United Nations report warns that the development of artificial intelligence could exacerbate global inequality and proposes a common framework for responsible AI development, as the adoption and investment in this technology is uneven across the world, UNN reports, citing The Guardian.
Details
"Access to AI tools alone does not provide equal benefits," the report states. "Countries that rely on foreign models, cloud infrastructure, and data transmission channels may gain access to AI while losing practical control over its standards, safeguards, and alignment with local conditions."
In a large-scale analysis by an independent international scientific panel on AI, established by the UN General Assembly last year as the "first global scientific body on AI," the risks and opportunities of AI are detailed — from transformative possibilities in agriculture and education to catastrophic consequences when malicious actors use AI to commit fraud and influence elections.
The preliminary report also functions as a toolkit, offering UN member states initial recommendations on ways to harness AI's potential for growth across various sectors while minimizing and mitigating threats. Proposals include developing local AI infrastructure such as data centers, improving AI literacy in schools and among the workforce, investing in developers, creating AI safety institutes, establishing strategies to combat disinformation, and continuously measuring the behavior of AI systems after release, "with real users, real tasks, and real environments."
While over a billion people now use AI weekly, access and types of use vary greatly around the world, "with adoption in the Global South significantly lagging behind the Global North," the report says. The US and China dominate the development of leading AI models, as well as investment in the computing infrastructure encompassing the hardware, memory, networks, and storage needed to run powerful AI models.
"The concentration of AI capabilities in a small number of firms and countries could lead to authoritarian capture and undermine democratic accountability," the report states.
The group advises countries lagging in AI development to consider significant investments in computing and information infrastructure. They note that attracting these funds requires ensuring a reliable energy supply and building data centers. However, the report acknowledges the environmental consequences of data centers, including their high energy and water consumption, as well as the potential for greenhouse gas emissions.
The authors also describe challenges in assessing the safety and ensuring oversight of increasingly powerful AI models.
The US lifts restrictions on powerful AI models, Anthropic reported01.07.26, 09:57
"Most countries, including many developed economies, lack the technical expertise to evaluate the most capable 'frontier' models or to meaningfully participate in their governance," they write.
The group of 40 independent scientific experts from around the world stated that this report is "the first of its kind." The UN, they argue, "is the leading global forum for cross-border risks of this magnitude" — and its approach is "scientific, not political."
Differences in languages and internet access exacerbate the digital divide.
"AI leaves most languages behind," the report notes. While generative AI tools work well in English and other widely used languages, "most languages are either excluded or have much lower performance."
These differences can have significant consequences, especially in the context of healthcare. The report cites an example of machine translation into Tigrinya, where smallpox is confused with syphilis, gonorrhea with diabetes, and the phrase "you were given intravenous antibiotics" with "you were given intravenous insecticides."
"These mistranslations can be life-threatening," the report states.
Some regions lack stable internet access, let alone the deployment of AI models. According to the International Telecommunication Union, over 2 billion people — nearly a third of the world's population — have no internet access at all.
New Chinese AI model raises concerns over hacking risks26.06.26, 05:40