Why the world needs a multilateral compact on AI, and why the Arab world cannot afford to wait

Talal Abu-Ghazaleh
April 2026 - Continued from part one

I recognize and acknowledge that AI is already doing much good for the world. Among its many applications, scientists are using it to accelerate drug discovery, climate researchers are building more accurate forecasting models and educators and farmers in underserved communities are accessing AI tools previously beyond their reach. These achievements deserve recognition as evidence of what this technology can be if guided by the full breadth of humanity’s needs.
The question I ask is not whether AI can produce benefits as it clearly can. The question is: who controls those benefits? Who shapes the conditions under which they flow? And who is structurally excluded from them? When the infrastructure of progress is owned by a handful of corporations, every achievement of AI becomes entangled with their platforms, their data agreements, and their strategic interests.
“The internet was once presented as the great equalizer. Instead, it became a system dominated by platforms whose algorithms now shape public opinion, mental health, and the outcomes of elections.”
The risks are not theoretical. Large-scale misinformation, AI-enabled cybercrime, social manipulation, and economic disruption are already present in the world, and they are exacerbated when decision-making authority is concentrated within a few systems owned by a handful of large tech firms. The values, priorities and agendas of such firms are inevitably embedded into these, and they affect everyone.
Throughout my years leading the UN Global Compact, I argued consistently that the governance of transformative technology must not be left to market forces alone. Markets are efficient engines of growth, but they are not equipped to make decisions that involve sovereignty, equity, and the long-term common good of humanity. AI governance requires what nuclear technology eventually required: binding international frameworks, transparency obligations, and mechanisms through which the nations most affected have a genuine voice.
I speak with particular urgency for the Arab world and the broader community of developing nations. If the infrastructure of AI is controlled by a small number of actors before sound governance frameworks are in place, the result will not be a level playing field. It will be a new form of digital dependency, in which nations must access intelligence itself through systems they do not control, on terms they did not negotiate, shaped by values that may not reflect their own. This is not a distant risk. It is the trajectory of AI as we speak.
“The capacity to reason, to learn, and to make decisions is a shared human resource, not a proprietary asset.”
What is required is not a rejection of AI. It is a refusal to accept that the concentration of AI power in private hands is either inevitable or acceptable. The world needs a new multilateral compact on artificial intelligence, one that establishes common standards for transparency, accountability, and equitable access. One that ensures developing nations are participants in AI governance, not merely consumers of systems designed elsewhere.
The first technological revolutions had changed what humanity could produce. This one is changing how humanity thinks, decides, and governs itself. That demands a response commensurate with its scale, not from individual companies exercising voluntary restraint, but from the international community acting with the seriousness and urgency this moment requires.
“The future of intelligence must not be left to those who own the infrastructure. It belongs to all of us.”