Chief executives from prominent artificial intelligence companies delivered stark warnings to the United Nations Security Council on Wednesday, urging the implementation of global oversight mechanisms to prevent catastrophic outcomes. Dario Amodei, CEO of Anthropic, cautioned that mismanaged AI could pose a threat to all of humanity, while Sam Altman of OpenAI echoed these sentiments, stating that civilization could “lose control of the future of AI” without proper governance.
The address took place during the annual UN General Assembly in New York City and was convened by France. Both Altman and Amodei emphasized that regulatory frameworks must not be dictated solely by technology hubs in San Francisco. “If AI is to be democratic, the most important decisions cannot be made by labs in San Francisco alone,” Altman asserted, calling for government accountability and democratic processes to shape the technology’s trajectory.
The meeting also highlighted the complex role of AI in cybersecurity. Clement Delangue, CEO of Hugging Face, recounted how his company was targeted by autonomous AI agents in recent attacks. Notably, he revealed that Hugging Face defended itself using a Chinese AI model, which faced fewer restrictions than comparable American tools. “We were attacked by AI, but more importantly, we defended ourselves with AI,” Delangue told the council, illustrating the dual-use dangers of the technology.
Despite the urgent calls for regulation from tech leaders and European officials, major powers such as the United States and China have been reluctant to embrace strict international controls. Michael Kratsios, representing the Trump administration at the Security Council, explicitly rejected the notion of centralized global governance. “We totally reject all efforts by international bodies to assert centralised control and global governance of AI,” he stated.
This resistance stems partly from the intense technological rivalry between Washington and Beijing. President Xi Jinping is expected to discuss AI regulation during an upcoming visit to Washington, but experts note that both nations are prioritizing rapid development and growth opportunities over collaborative restraint. Daniel Forti of the International Crisis Group observed that while member states recognize the need for cooperation, the largest players remain focused on competitive advantages.
The UN has been working to establish norms for several years, adopting its first resolution on AI in 2024, which called for protecting personal data and human rights. Last year, the General Assembly established two new bodies: the Independent International Scientific Panel on AI and the Global Dialogue on AI Governance. Yoshua Bengio, co-chair of the scientific panel, told the Security Council that the dangers are “real and imminent,” adding that the threat does not respect borders and requires collective action beyond the capacity of any single nation.
If the threats truly don’t respect borders, why are superpowers still treating AI like another arm of geopolitical competition rather than a shared risk?
America rejecting global oversight while building the most powerful tools? This doesn’t scream ‘responsible innovation’ to me at all.
It is fascinating that Hugging Face used a Chinese model to defend against AI attacks. The irony is not lost on anyone watching this closely.