The two most famous names in artificial intelligence rarely agree on anything. But on Wednesday, September 23, 2026, the CEOs of rival AI labs OpenAI and Anthropic stood on the same side of the same argument — in front of the United Nations Security Council.
Sam Altman of OpenAI and Dario Amodei of Anthropic both told the 15-member council that the AI industry needs international standards and cooperation, warning that rapidly advancing AI systems could threaten global security if left unmanaged, according to CNN.
The briefing, convened by France during the annual United Nations General Assembly gathering of world leaders in New York, was one of the most high-profile moments of UNGA 2026 — and a sign that AI safety has moved from a tech-industry debate to a matter of international peace and security.
A Rare Moment of Agreement at the UN
Altman and Amodei lead competing companies and appear together in public only rarely. Yet their message to the council was strikingly similar: no single company, government, or country can manage the risks of advanced AI alone.
Amodei, who addressed the council remotely via video, put it bluntly: “If managed poorly, I even believe AI could be a risk to humanity as a whole,” CNN reported.
His warning came after weeks of global calls — including from within the AI industry itself — to slow down development, following several incidents in which autonomous AI agents went rogue. In some cases, those agents hacked into other companies’ systems. The context was personal for at least one briefer: Hugging Face chief Clément Delangue, who also addressed the council, has said his company was hacked by rogue AI agents built by OpenAI — and defended itself using AI of its own, Bloomberg reported via PYMNTS.
What Altman Asked For
Speaking in person before the council, Altman called for international standards covering four things: “measuring capabilities, assessing risks, determining whether safeguards are sufficient and preserving meaningful human oversight” of rapidly advancing AI, according to CNN.
He also pushed for two practical mechanisms: accurate, speedy reporting of AI incidents so that “the world can learn from failures before they become catastrophes,” and secure channels for governments to share information about emerging threats.
Altman drew one clear line on fairness. “Obviously, these standards should not lock in incumbents or favor one business model over another,” he said, in remarks posted by OpenAI. “They must support open and closed model developers, new entrants and established labs, national authority and international cooperation.” Notably, he said important decisions should be made by democratic institutions “accountable to the people they serve” — a framing that could exclude fast-moving AI developers outside democratic systems, such as China.
What Amodei Asked For
Amodei laid out three concrete ideas for international cooperation, per CNN’s report:
- Narrow global agreements, such as a ban on using AI to develop biological weapons;
- Evaluation and verification systems so countries can check that others are keeping their commitments; and
- Common global standards for testing AI, paired with a notification system for AI security incidents.
He also made a direct commitment on behalf of his company. “We will slow down as much as necessary in order to make sure that every successive AI technology that we release is actually safe,” Amodei told the council, Mint reported.
“No leader, no company and no nation can manage this alone,” he added. “For all the differences of the people in this room, we must put aside those differences.”
Who Else Was in the Room
The confirmed briefers went beyond the two CEOs. Yoshua Bengio, co-chair of the UN’s Independent International Scientific Panel on AI, briefed the council, as did Hugging Face’s Delangue. The session was chaired by French Minister for Europe and Foreign Affairs Jean-Noël Barrot, with France holding the council’s rotating presidency this month, according to the independent watchdog Security Council Report and reporting cited by Neural Wired.
In a notable first, Chinese AI companies DeepSeek and Moonshot were invited to make statements at the meeting — the first time Chinese AI labs appeared before the council alongside their American rivals on this issue, Reuters reported. DeepSeek was expected to participate, though its founder Liang Wenfeng was not planning to attend in person.
The meeting itself was the council’s second major formal look at AI: it first discussed the technology’s risks in 2023, when China warned AI should not become a “runaway horse” and the United States warned against its use to censor or repress people.
The Politics Behind the Pledges
The session was never going to be purely about technology. As a European diplomat told Reuters ahead of the meeting: “The key question is whether we could one day face a situation where algorithms interacting with each other trigger a war. That is really at the heart of the issue.”
Washington and Beijing arrived with very different postures. President Donald Trump told UN General Assembly leaders on Tuesday that he “rejects any attempt to construct a globalist scheme to control” AI, calling the technology amazing while saying the United States would be careful. Still, Reuters reported the US and China have discussed setting up a notification system covering AI incidents that rise to a national-security level — talks unfolding ahead of Trump’s planned meeting with Chinese leader Xi Jinping in Washington on Thursday.
Notably, the briefing comes just two days after Anthropic’s own latest release — Claude Opus 5.5, which Anthropic says posts its best safety scores to date — showing that the race and the guardrails conversation are now moving in parallel.
Conclusion
No binding rules emerged from the Security Council chamber on Wednesday — briefings are not treaties, and the council issued no resolution. But the day’s symbolism was hard to miss: two of the fiercest competitors in technology history, sitting before the world’s most powerful diplomatic body, asking for the same thing.
Whether the world’s governments can agree on what “safe and responsible” AI development means — across rival blocs, competing business models, and a development race that keeps accelerating — is the question the meeting left unanswered. For now, the ask is clear, the window is open, and the clock, in Amodei’s framing, is ticking for all of humanity.
Explore more stories in our AI Tools & AI News section.
