By Sadaf Sundas Riaz — SCN News
UNITED NATIONS — Pakistan called on the international community on Thursday to keep humans firmly in control of artificial intelligence and prevent the technology from becoming another arena of military competition, warning the UN Security Council that the same safety rules intended to contain AI risks must not become a mechanism for excluding developing countries from one of the defining technologies of the century.
Deputy Prime Minister and Foreign Minister Ishaq Dar told a high-level Security Council briefing on Artificial Intelligence and International Security that AI was already reshaping societies, economies and the security environment. Describing it as both a dual-use technology and a potential force multiplier in modern warfare, Dar said international action was needed before increasingly capable systems compressed military decision-making and increased the risk of miscalculation during crises.
His central message was that responsibility for consequential decisions cannot simply be transferred from people to machines. Pakistan wants human judgement and control retained throughout the design, development and deployment of AI systems, particularly where their use could affect international peace and security. Dar said governments must also retain the ability to determine the structures within which AI operates and the limits placed on the speed and nature of future capabilities.
“We must ensure that AI does not become another facet of the ongoing arms race,” Dar told the Council, calling for guardrails, regulation and governance frameworks alongside restraint and confidence-building measures in the military domain. Pakistan argues that increasingly automated decision-making could shorten the time available to governments during a confrontation, increasing the danger that erroneous information, system failures or misinterpretation could contribute to escalation.
The intervention came as AI moved unusually high onto the international-security agenda during UNGA81. France convened the Security Council meeting with participation from major figures in the technology sector, including OpenAI CEO Sam Altman, Anthropic CEO Dario Amodei and Hugging Face CEO Clément Delangue, alongside Yoshua Bengio, co-chair of the UN Independent International Scientific Panel on AI. The discussion brought governments and developers of frontier AI systems into the same forum as concerns grow over increasingly powerful models and their potential military and security applications.
Pakistan’s intervention, however, went beyond the emerging debate over whether advanced AI could eventually escape meaningful human control. Dar highlighted more immediate threats already associated with the technology, including AI-generated misinformation, fake news, cyberattacks and attacks against critical infrastructure by state and non-state actors. Those capabilities, he argued, mean AI governance is increasingly connected to international security rather than being solely a question of technology regulation.
The military dimension is particularly important to Islamabad. Pakistan has for years advocated international regulation of emerging military technologies and has called for a legally binding instrument addressing fully autonomous weapons systems. Its established position is that weapons capable of operating without effective human control create difficult questions over accountability, predictability, unintended consequences and compliance with international humanitarian law.
But Dar paired that security argument with a second warning that gives Pakistan’s position a distinctly developing-country dimension: AI safety cannot become technological exclusion.
He cautioned that if a small group of technologically advanced states or corporations is allowed to determine both the pace of AI development and who can access the technology, attempts to make AI safer could deepen the technological divide between wealthy and developing countries. Restrictions designed in the name of security, Pakistan argues, should not leave countries outside the major AI powers permanently dependent on systems and standards created elsewhere.
That argument places Pakistan within a wider dispute increasingly visible between major technological powers and developing economies. Advanced chips, computing capacity, frontier AI models and the infrastructure required to train them are concentrated among a relatively small number of countries and companies. At the same time, national-security concerns are driving governments to impose controls on the transfer of some of the technologies that make advanced AI possible.
Dar said the risks created by uncontrolled AI would ultimately be shared internationally and therefore decisions over managing those risks could not be left to only a handful of companies or capitals. Pakistan wants governance negotiations conducted through open and representative international forums where developing countries participate in writing the rules rather than simply implementing standards established by technologically dominant states.
The debate has become increasingly important because there is still no single global authority regulating artificial intelligence. Different governments are developing national rules while technology companies establish their own safeguards and technical standards, raising the possibility of a fragmented international system just as AI capabilities become more globally consequential.
UN Secretary-General António Guterres has used his final General Assembly session to press for stronger multilateral oversight. In his UNGA81 address this week, he called for an international framework to manage AI risks and independent mechanisms capable of monitoring rapidly developing technologies, arguing that international governance has not kept pace with technological change.
The Security Council debate demonstrated that agreement on the existence of AI risks does not necessarily translate into agreement over how to regulate them. The United States has resisted approaches that could unnecessarily slow innovation or create governance structures Washington considers overly restrictive, while China has argued that technological restrictions can preserve advantages held by established powers. Pakistan’s intervention sought to bridge those competing concerns by supporting safeguards while insisting they be developed multilaterally and applied without locking poorer countries out of AI development.
The companies developing some of the world’s most advanced systems are themselves calling for stronger cooperation on certain risks. OpenAI has urged coordinated international technical standards, common methods for evaluating increasingly capable models and mechanisms for reporting serious AI incidents. Its proposals come as researchers examine whether future systems capable of improving aspects of their own performance could introduce risks that national regulation alone would struggle to manage.
Anthropic CEO Dario Amodei told the Security Council that poorly managed AI could eventually pose a risk to humanity, while industry representatives also discussed the need for safeguards as capabilities advance. Those warnings have intensified a debate over whether governments should deliberately slow the development of the most powerful systems or instead concentrate on safety standards without restricting technological progress.
Pakistan did not call for stopping AI development. Its position instead combines three elements: human control over consequential systems, international safeguards against militarisation and equitable access to the technology. That allows Islamabad to support stronger governance while resisting a regulatory architecture it believes could consolidate AI power among a small number of governments and corporations.
Dar also welcomed the first UN Global Dialogue on AI Governance held in Geneva this year, presenting it as a potential foundation for broader international participation. Pakistan wants the United Nations to remain central to the governance process rather than allowing competing national regulatory systems or technology companies alone to determine global standards.
Islamabad has simultaneously aligned itself with another emerging international initiative. Pakistan is a founding participant in China’s proposed World Artificial Intelligence Cooperation Organization, or WAICO, which Beijing has presented as a mechanism for international cooperation and broader access to AI. Dar described the initiative as supporting development of artificial intelligence “for good, and for all.”
That participation also reflects the geopolitical competition increasingly surrounding AI governance. Washington and Beijing are competing for leadership in advanced models, semiconductors and computing infrastructure while simultaneously discussing whether some common safeguards are necessary to prevent dangerous incidents. AI is consequently becoming both an area of strategic rivalry and one where cooperation between competing powers may eventually be unavoidable.
For Pakistan, that combination makes inclusive governance particularly important. Countries without the computing resources or technology companies of the United States and China could otherwise face rules and technological dependencies shaped almost entirely by the rivalry between the two AI powers.
Dar therefore urged confidence-building measures not only around military applications but also around increasingly powerful foundational AI systems. Such mechanisms could include communication between governments, greater transparency around serious incidents and agreed safeguards intended to prevent misunderstandings or unintended escalation as AI becomes more deeply integrated into national-security systems.
The intervention also connects AI governance with Pakistan’s broader arms-control policy. Islamabad has previously argued that unregulated military applications of emerging technologies — including autonomous weapons, armed drones and other AI-enabled systems — could lower the threshold for conflict and create accountability gaps when machines become increasingly involved in decisions traditionally made by humans.
Thursday’s Security Council debate showed that the international argument is now moving beyond whether AI represents a security issue. The harder questions concern who writes the rules, how much control humans must retain, how military applications should be constrained and whether safety restrictions can be imposed without concentrating technological power among a few states and companies.
Pakistan’s position at UNGA81 was that those questions cannot be separated. Islamabad supports stronger safeguards against dangerous AI and its military use, but argues that global governance will lack legitimacy if the countries most vulnerable to technological exclusion have little influence over the rules.
That leaves Pakistan advocating a balance likely to become increasingly difficult as AI capabilities accelerate: keep humans in control, prevent artificial intelligence from becoming another global arms race, but ensure that protecting the world from AI risks does not become an excuse for denying the developing world access to AI itself.