An Independent International Scientific Panel on Artificial Intelligence established by the United Nations General Assembly has warned that although artificial intelligence (AI) has the potential to drive unprecedented advances in science, healthcare, education, agriculture, and economic productivity, its benefits will not be shared equitably unless governments invest in infrastructure, human capacity, governance, and international cooperation to address widening global inequalities and emerging risks.
In a 59-page report titled “Preliminary Report of the Independent International Scientific Panel on AI: Evidence-based assessment of opportunities, risks, and impacts of AI” released on July 1, 2026, the Panel presented a comprehensive scientific assessment of the state of AI, its rapid evolution, its societal implications, and the policy choices needed to ensure that the technology serves humanity.
The 40-member Panel, established by the UN General Assembly in its resolution 79/325 of 2025, is co-chaired by Ms Maria Ressa, a journalist, co-Founder and CEO of Rappler, an online news website based in Manilla, a Nobel Peace Prize laureate and a dual citizen of the Philippines and the United States; and Professor Yoshua Bengio, a Canadian computer scientist and Professor of Computer Science at the University of Montreal in Canada and co-President and Scientific Director of the non-profit organization, LawZero.
The Panel, composed of independent scientists and experts from all five UN regions, also has, as a member, Professor Rita Orji, a Nigerian computer scientist who serves, among other capacities, as Canada Research Chair in Persuasive Technology at Dalhousie University in Halifax, Canada, and Director of the University’s Persuasive Computing Lab.
In a joint message upon the release of the report, the co-chairs expressed deep concern over the race for competitive advantage, saying it is “leading the world into risks whose magnitude is grossly underestimated.” They also raised alarm that “The most consequential decisions about humanity’s future are being made by a handful of companies and a handful of governments.”
According to them, “AI is sold now as a commodity, but it may not stay that way. It is becoming a weapon and a tool of domination. Concentration of capability is becoming concentration of political power. The sovereignty of most nations is at stake.”
The Preliminary Report noted that AI is advancing at an unprecedented pace, moving beyond systems that merely generate text, images, and code towards increasingly autonomous “agentic” systems capable of reasoning, making decisions, using digital tools, and carrying out complex tasks with diminishing human oversight.
It said these developments represent a fundamental shift in the nature of AI and require governance frameworks that are capable of keeping pace with technological change.
The report also noted that AI capabilities continue to improve rapidly while investments in computing infrastructure and frontier AI development have reached levels previously associated only with major national industrial programmes.
However, it warned that the benefits of these investments are becoming increasingly concentrated in a small number of companies and countries, raising concerns about market concentration, geopolitical influence, and unequal access to AI resources.
The Panel expressed particular concern about the widening AI divide between developed and developing countries, arguing that the divide extends beyond access to AI technologies and encompasses the capacity to build computing infrastructure, develop skilled talent, regulate AI effectively, and influence international governance.
It noted that 118 countries, predominantly in the Global South, are not actively participating in major AI governance discussions, while fewer than one-third of developing countries have adopted national AI strategies.
The report also highlighted the limited linguistic and cultural diversity represented in current AI systems, noting that although more than 7,000 languages are spoken globally, AI models currently support only a small fraction of them, leaving much of the world’s population underrepresented in AI development and evaluation.
The Panel calls for greater investment in datasets, benchmarks, and AI systems that reflect diverse languages and cultural contexts.
Despite these concerns, the report emphasized AI’s significant potential to improve lives when deployed responsibly and cites successful applications in healthcare, including AI-assisted disease screening and clinical support systems; improvements in scientific research; advances in agricultural forecasting and biodiversity conservation; and growing use of AI to support teaching and learning.
However, it stressed that these benefits depend on appropriate infrastructure, trained professionals, supportive institutions, and effective governance rather than on technology alone.
The Panel warned that the emergence of agentic AI substantially changes the risk landscape, as systems capable of acting autonomously, coordinating with other AI systems, and executing tasks in real-world environments introduce new challenges relating to cybersecurity, operational control, accountability, and human oversight.
In addition, it said, existing governance mechanisms were designed for earlier generations of AI and are not yet equipped to manage increasingly autonomous systems.
Among the report’s most significant findings is its warning that AI is increasingly threatening information integrity and democratic discourse as a result of the ease with which AI systems can generate convincing text, images, audio, and video, which has accelerated the spread of disinformation and deepfakes, making it progressively harder for citizens to distinguish authentic information from manipulated content.
The report cautioned that AI-generated persuasion technologies, synthetic consensus, and the so-called “liar’s dividend”, where genuine evidence can be dismissed as fabricated, pose growing risks to elections, public trust, and democratic institutions.
It further identifies significant human rights concerns arising from AI, including threats to privacy through large-scale surveillance, discriminatory outcomes caused by biased algorithms, online violence against women, and increasing risks to children through AI-generated sexual abuse material and manipulative AI companions.
The report called on governments and developers to integrate human rights considerations throughout the AI lifecycle using impact assessments, transparency requirements, and accountability mechanisms.
Recognising the growing importance of public understanding of AI, the Panel argued that AI literacy should become a priority for governments, educational institutions, and employers, noting that existing AI literacy initiatives remain limited and should be expanded to include not only technical knowledge but also ethical, legal, and societal dimensions of AI.
It insisted that AI literacy should complement and not replace the developer’s responsibility and regulatory oversight.
The Panel concluded that governing AI under conditions of uncertainty will require international cooperation, stronger public institutions, and sustained investment in scientific capacity, particularly in developing countries.
Although it acknowledged that AI offers extraordinary opportunities for sustainable development, scientific discovery, and economic growth, it warned that these outcomes are not inevitable.
Whether AI ultimately benefits humanity, the report argued, will depend on deliberate policy choices, effective governance, and ensuring that countries around the world have the capacity not only to use AI, but also to shape its future.
The final report is expected to be published later in the year.
