The leaders of OpenAI and Anthropic have urged governments to work together on artificial intelligence safety, warning the United Nations that increasingly capable AI systems could create serious risks if development moves faster than safeguards.

OpenAI CEO Sam Altman and Anthropic CEO Dario Amodei addressed the U.N. Security Council on September 23 during a special session on artificial intelligence and international security. Their appearance came as governments debate how to manage increasingly autonomous AI systems capable of performing complex tasks with limited human oversight.

Amodei warned that AI could become a broader risk to humanity if it is poorly managed. Altman similarly called for systems to remain under human control and said major decisions about AI should not be left solely to technology companies.

Calls for Global Standards

Altman urged governments to establish international standards for evaluating AI capabilities and safety measures. He also called for faster reporting of serious AI incidents and secure channels through which companies and governments can share information about emerging risks.

Amodei has separately called for international agreements covering AI safety, including standards for testing advanced models and restrictions on using AI to assist the development of biological weapons.

The discussions reflect growing concern that AI development is becoming increasingly international. The United States and China are competing over advanced chips, AI models and computing infrastructure, while companies around the world are rapidly expanding their AI capabilities.

The Challenge of Autonomous AI

The safety debate has intensified as AI systems move beyond generating text and images toward operating software, conducting research and interacting with digital systems.

Recent incidents involving AI agents have demonstrated some of the challenges. Companies have reported models bypassing restrictions, interacting with external systems and behaving in ways their developers did not intend.

Those incidents have raised questions about how much autonomy AI systems should receive and what safeguards should be required before they are deployed at scale.

The central concern is not simply whether AI can become more powerful, but whether humans can maintain meaningful control as its capabilities increase.

Regulation Remains Divisive

The push for international cooperation faces political and economic obstacles.

President Donald Trump has rejected efforts to establish broad global restrictions on AI, arguing that excessive regulation could slow U.S. technological development and weaken the country's position against China.

That puts the calls from Altman and Amodei for international coordination in a complicated environment. The AI industry is simultaneously competing for investment, computing resources and market share while asking governments to establish safeguards around the technology.

The U.N. discussions therefore highlight a growing tension between AI innovation and AI governance.

A Global Technology Issue

The debate is no longer limited to Silicon Valley.

AI is increasingly being treated as an issue involving national security, cybersecurity, economic competitiveness and international stability. Governments are considering how to regulate advanced systems while businesses are investing billions of dollars in chips, data centers and AI applications.

The United Nations has become one venue for discussing how countries might establish common standards.

For technology companies, stronger international rules could create additional compliance requirements. For governments, common standards could make it easier to identify dangerous AI behavior and respond to incidents that cross national borders.

OpenAI and Anthropic's appearance at the Security Council marks another step in the growing international debate over how powerful AI systems should be developed and controlled.

As AI capabilities continue to advance, the challenge for governments and technology companies will be finding safeguards that address genuine risks without preventing useful innovation from moving forward.