Altman and Amodei urge UN for global AI safety cooperation

The direct engagement of private technology executives with the UN Security Council represents a departure from traditional state-centric security diplomacy. Historically, international arms control and dual-use technology regimes have been negotiated exclusively between sovereign states through treaties and export control arrangements like the Wassenaar Arrangement. The involvement of corporate leaders as primary advocates for binding restrictions mirrors precedents seen in nuclear non-proliferation, where scientific communities occasionally bypassed governments to warn of existential risks, though rarely with such direct access to security apparatuses. This pattern reflects a growing recognition in international relations theory that non-state actors possess critical technical expertise necessary for governing emerging technologies. Previous AI governance efforts, such as the UNESCO Recommendation on the Ethics of Artificial Intelligence, relied on consensus-based soft law. The shift toward seeking Security Council intervention suggests a belief that voluntary frameworks and soft law have reached saturation, necessitating harder security-focused mechanisms typically reserved for threats to international peace and stability.

On September 23, 2026, Sam Altman of OpenAI and Dario Amodei of Anthropic jointly addressed the United Nations Security Council (UNSC) to advocate for a formalized international framework governing artificial intelligence safety. The appearance marked a significant escalation in industry-led diplomacy, with both CEOs arguing that voluntary national measures are insufficient to manage catastrophic risks associated with advanced AI systems. According to CNN, Altman specifically urged the council to create international standards for "measuring capabilities, assessing risks, determining whether safeguards are sufficient." This appeal was delivered against a backdrop of geopolitical friction; CNBC reported that Altman’s push for UN cooperation followed a rebuff from the Trump administration regarding domestic coordination, suggesting the industry is seeking multilateral venues as alternative pathways for governance.

The logistical execution of the appeal highlighted the distinct approaches of the two leaders. As documented on Bluesky by user wildebees, Sam Altman appeared in person before the Security Council, while Dario Amodei participated via video link. Despite the difference in physical presence, their messaging was aligned. Amodei utilized the broader UN General Assembly platform to call for "narrow" international agreements focused specifically on safety, distinguishing technical risk mitigation from broader political disputes. This distinction was echoed in Mexican media coverage; El Financiero reported on September 25 that the CEOs’ primary request was for global cooperation to avoid risks, framing the debate as a technical necessity rather than a purely diplomatic exercise.

The timing of this joint appeal coincided with parallel geopolitical overtures. On the same day as the UNSC briefing, China’s President Xi Jinping publicly urged the United States to cooperate on AI, according to CNBC. This convergence created a complex diplomatic environment where private sector leaders and rival superpowers were simultaneously signaling a desire for engagement, albeit potentially with divergent underlying motives. ABC News noted that both Altman and Amodei stressed the need for global cooperation to set safety standards as the technology rapidly develops, reinforcing the narrative that the pace of innovation has outstripped existing regulatory containment.

Al Jazeera reported that the CEOs told the 15-member council that humanity could face peril absent global regulation, explicitly framing the issue as a crisis requiring the UNSC’s mandate. The New York Times corroborated this, stating that America’s AI leaders warned the UN of possible peril without a coordinated response to ensure AI remains under human control. However, despite the high-profile nature of the testimony, the United Nations has not yet committed to creating dedicated AI safety governance structures. The event currently stands as a record of industry concern and proposed frameworks rather than an enacted policy shift. Mexico Business News summarized the outcome as an urging of standards, reflecting the non-binding nature of the CEOs' recommendations at this stage.

The sequence of events suggests a calculated strategy by leading AI labs to internationalize safety standards. By bringing their concerns directly to the UNSC, Altman and Amodei have attempted to bypass stalled or hostile domestic regulatory environments. The emphasis on "measuring capabilities" and "narrow" agreements indicates a preference for technically verifiable compliance mechanisms over broad, aspirational principles. Whether this appeal will translate into binding resolutions remains unconfirmed, but the act of testifying before the Security Council establishes a new precedent for private sector involvement in high-level international security discourse.

OpenAI and Anthropic risk reputational capital by aligning with UN processes that may conflict with U.S. sovereignty interests. If successful, this could establish binding international verification regimes affecting all frontier labs. Failure to secure UN action after such a high-profile appeal would weaken the industry's claim to self-regulatory competence. The magnitude involves potential restructuring of global AI compliance architecture, impacting R&D timelines and cross-border deployment. Geopolitically, the move intersects with U.S.-China AI competition, where unilateral Western standards could fragment the global ecosystem. Domestic U.S. policymakers may view this as circumvention, risking retaliatory legislation. Conversely, Global South nations may leverage this precedent to demand greater influence over AI governance previously dominated by Western tech firms and governments.