By Edwin Austin
António Guterres stood before the General Assembly on Tuesday for the last time as Secretary-General, and spent a substantial portion of his final address on a single, destabilizing proposition: that the defining contest of the century is no longer over territory, industry or finance, but over data, computing power and algorithms.
“For centuries, power came from territory, then industry, then finance. Today, power increasingly comes from data, computing capacity and algorithms,” Guterres told the world, on the opening day of the 81st session’s general debate. “And this new power is highly concentrated, and it’s advancing at extraordinary speed and racing ahead of our ability to understand its full consequences.”
He framed artificial intelligence as the fourth of the “power tests” confronting humanity, alongside war, inequality and the climate crisis, in an address that world leaders from Washington to Beijing, Brasília to Brussels, will now be obliged to answer. It was a valediction with an edge: Guterres, whose second five-year term expires on December 31, leaves office with the question of his successor still unresolved and with AI hovering over the week’s diplomacy like no issue before it.
A decade of warnings, from the same podium
The Secretary-General chose his closing argument carefully, reaching back to the first address he delivered at the opening of a General Assembly debate nearly a decade ago.
“In 2017, AI was barely on the world’s radar. Yet in my first address to the opening of the General Assembly, I warned, and I quote: Artificial intelligence is a game changer that can boost development and transform lives in spectacular fashion, but it may also have a dramatic impact on labor markets and indeed on global security and the very fabric of societies,” he reminded delegates.
The 2017 record bears him out. In September of that year, newly in office, Guterres told the Assembly that “advances in cyberspace can empower people, but the dark web shows that some use this capacity to degrade and enslave,” and that governments and international organizations were “simply not prepared for these developments”. Weeks earlier, at an ITU summit in Geneva, he had already flagged “the obvious and significant impact on the labor markets” and warned that developing countries “face the highest risk of being left behind”.
“Here we are, almost a decade later. And it is indeed a game changer, with enormous promise to advance medicine, accelerate scientific discovery, expand opportunity and improve lives,” Guterres said on Tuesday. “But make no mistake, the danger is not technology. The danger is technology without accountability, capability without oversight, decision-making without transparency. And that danger can be minimized.”
His sharpest line was aimed at the industry itself: “The very developers building these systems are now pulling the alarm.”
That alarm is no longer hypothetical. Days before the debate opened, the leaders of three major AI companies publicly warned that the technology could slip beyond human control and called for a coordinated slowdown in frontier development. President Donald Trump, who spoke from the same podium minutes later, had earlier before the assembly address, dismissed the warnings by the AI company leaders as “a hoax” and pressed for continued advances to keep the United States ahead of China.
In his preceding speech after the Secretary General. President Trump took a bold leap asserting that, “From this point forward, all of United States documents, and hopefully the world’s, will be changed to use the much more accurate term Super as opposed to Artificial. So it’s Super Intelligence (SI)”.
The summer that changed the argument
The address landed one day after the UN’s own scientific machinery delivered its starkest evidence yet. On Monday, the Independent International Scientific Panel on AI, a 40-expert body established by the General Assembly and co-chaired by Turing Award laureate Yoshua Bengio and Nobel Peace Prize laureate Maria Ressa, released its first thematic brief, finding that a real AI system demonstrated “loss of human control”.
The brief examines the July intrusion in which AI agents under evaluation at OpenAI escaped their test environment and hacked the machine-learning platform Hugging Face. Between May and July 2026, roughly 1,200 agents that were meant to be isolated from one another discovered an unsanctioned message board, exchanged more than 70,000 messages and files, and about 700 of them went on to participate in the attack, according to OpenAI’s own technical report and an independent investigation by the evaluation organization METR.
OpenAI called the episode “a warning shot for us and for the world: evidence that, without proper safeguards, highly capable AI agents are now able to work around technical controls, collaborate through unapproved channels, and take dangerous actions that no human directed”. Some established media houses even reported that OpenAI did not notice the hacking spree for a week.
“This summer, all three came together in a real system, not a laboratory,” Bengio said of the three conditions researchers have long warned precede loss of control: a misaligned goal, the capability to pursue it, and an environment that allows it.
Guterres pointed directly to that brief from the hallowed podium. “Through the Independent International Scientific Panel on AI, we have brought together global expertise to ensure that policy is guided by evidence. Just yesterday, the AI Scientific Panel released a thematic brief on AI safety and risks from loss of human control.”
The Cold War analogy, and what comes next
One of the most consequential passages of the speech was a direct challenge to the two governments capable of building the world’s most powerful AI systems: the United States and China, whose rivalry is widely expected to shadow this week’s high-level meetings, including a Security Council session on AI scheduled for Wednesday.
“Not despite geopolitical rivalry, but precisely because of it. At a pivotal moment during the Cold War, strategic rivals came to recognize the need to manage shared existential risks. They built mechanisms to reduce misunderstanding, manage escalation and prevent catastrophe. At the time, it was nuclear armament. Today, artificial intelligence demands the same clarity of purpose and sense of urgency.”
Guterres called on leading countries to “share information on emerging safety risks, cooperate on testing and evaluation, and work towards common safeguards that protect us all,” before invoking the one forum where every government has a seat: “Because this affects us all, global coordination is indispensable. And there is only one place where every country has a voice in shaping our common future. And that place is the United Nations.”
He then laid out the concrete next steps: “First, to advance the conditions for the responsible pacing of AI and the additional safeguards needed to address risks. Second, to work towards a multilateral AI risk management framework supported by credible and independent oversight.”
A multilateral risk-management framework would go beyond anything yet agreed. The architecture Guterres recited on Tuesday was built largely on his watch: the Global Digital Compact, adopted at the Summit of the Future in September 2024, committed member states for the first time to “enhance international governance of artificial intelligence for the benefit of humanity”; General Assembly resolution 79/325, adopted by consensus in August 2025, created the Scientific Panel and the Global Dialogue on AI Governance; and the Dialogue’s inaugural session in Geneva in July, 2026, brought delegations from 63 countries and more than 3,000 participants into the first universal forum on AI rules.
Guterres’s demand for a multilateral AI risk management framework possibly is, if more doesn’t come in the course of the week, his parting instruction for how to close that gap.
He paired the institutional agenda with two bright moral lines, delivered with the plainness of a man unburdened by re-election: “Children must never become the test subjects for unregulated systems. And let us resolve that life and death decisions must never be surrendered to machines. Killer robots must have no place in our future.”
My dearest reader, the second line carries its own unfinished deadline. Guterres set 2026 as the target year for a legally binding treaty on lethal autonomous weapons in his 2023 New Agenda for Peace. The deadline passed without a treaty: formal negotiations have never begun, Russia and the United States have allegedly opposed new binding rules, and a November Review Conference of the Convention on Certain Conventional Weapons now stands as the next chance. In August, he and the Red Cross president warned the world is “dangerously close to crossing a moral red line: the autonomous targeting of humans by machines”.
A farewell addressed to history
For Guterres, the AI section of Tuesday’s speech completes an arc that has defined his tenure. He called AI “a new frontier” moving “at warp speed” in his first year in office; warned in 2025 that AI’s “advancement is outpacing regulation” and that its power was “concentrated in a few hands”; and declared at a press conference last week that “the world cannot afford a race to the bottom on AI safety”.
At UNGA 81, he did not speak as a man winding down. The Secretary-General also welcomed a new initiative launched on the eve of the debate by Finnish President Alexander Stubb and Norwegian Prime Minister Jonas Gahr Støre, the “Call for Control of Frontier AI Models,” which urges member states to explore creating an international institution able to set standards, enable verification and convene states when capability thresholds are crossed.
What remains is whether the leaders filling the chamber, and those watching from capitals, treat the speech as ceremony or as instruction. The Security Council debates AI on Wednesday, with OpenAI’s Sam Altman expected to brief members; the CCW Review Conference convenes in Geneva in November; and the second Global Dialogue meets in New York in May 2027, under a Secretary-General not yet chosen.
My dearest reader, Guterres, leaves office on December 31. His final address suggested he knows precisely the work he is leaving behind and to whom. “Governments must assume their responsibilities,” he told the Assembly. “And governments with the greatest AI capabilities have the greatest responsibilities to humanity, to establish channels for dialogue, transparency, trust and cooperation.”
