• OpenAI CEO Sam Altman will address the UN Security Council today, advocating for international coordination on AI safety without backing a specific global regulator.
  • The meeting, convened by France, highlights the security risks of frontier AI models and includes participation from Anthropic CEO Dario Amodei and other industry leaders.
  • OpenAI's position favors common baseline standards and democratic oversight to avoid fragmented regulations that could hinder innovation.

Altman to Make Case for Global AI Safety Standards

Sam Altman, CEO of OpenAI, is set to speak before the United Nations Security Council today, September 23, in a session convened by France to address the intersection of artificial intelligence and international security. According to people familiar with the matter, Altman will argue for international coordination on AI safety, promoting common baseline standards and democratic oversight, while cautioning against rules that could unduly slow beneficial innovation.

The meeting, held under the Security Council's "Maintenance of international peace and security" agenda and chaired by French Foreign Minister Jean-Noël Barrot, comes amid growing concerns over the malicious use of frontier AI models, risks to critical infrastructure, and the potential loss of human control over increasingly autonomous systems. France, which holds the Council's rotating presidency, aims to elevate AI governance to a security imperative.

Altman's participation underscores OpenAI's push to shape global AI policy. The company, which announced a $122 billion financing round in March 2026 at an $852 billion post-money valuation, has been vocal about the need for safety measures. However, OpenAI is not endorsing a specific international regulator, instead favoring flexible coordination that preserves innovation. "We believe that common safety standards and democratic oversight are essential, but a one-size-fits-all global regulator could stifle progress," a source close to the company said.

Anthropic CEO Dario Amodei is expected to join remotely, along with Hugging Face CEO Clément Delangue and UN AI-panel co-chair Yoshua Bengio. Notably, Chinese AI firms, including DeepSeek and Moonshot, have been invited to make statements—a sign that any credible framework must engage the United States, China, the EU, and other major AI powers. Their participation, even if limited, could pave the way for discussions on shared minimum standards, such as evaluation practices, incident reporting, and cybersecurity safeguards.

The Security Council briefing follows a recent public convergence among frontier-AI leaders. Amodei has advocated for slowing the pace of capability gains and allowing more outside scrutiny of advanced labs, while Altman and others have voiced support for stronger safety measures. This alignment reflects mounting pressure on the industry to address risks without sacrificing competitiveness.

Regulatory Patchwork Drives Call for Coordination

The backdrop to today's meeting is a fragmented global regulatory landscape. The European Union's AI Act, with obligations for general-purpose AI models that began applying in August 2025 and broader enforcement from August 2026, has set a high bar for compliance. In contrast, the United States lacks a comprehensive federal AI statute, leaving a patchwork of state-level rules. This divergence creates uncertainty for companies operating globally and underscores the appeal of common international standards.

OpenAI's own trajectory illustrates the stakes. The company, now structured as a public-benefit corporation controlled by the nonprofit OpenAI Foundation, projects 2026 revenue of roughly $36 billion, rising to $350 billion by 2030, according to an internal presentation reported by Reuters. Yet it also expects a cumulative cash burn of about $278 billion from 2026 through 2030, with computing and infrastructure spending totaling around $856 billion by 2030. These forecasts highlight the immense capital requirements of frontier AI and the need for a stable regulatory environment to attract investment.

"Regulatory stability is crucial for long-term planning," said Andrea Valeri, Blackstone (BX)'s country chairman for Italy, at a recent conference. "Fragmented rules increase compliance costs and slow deployment." His comments echo wider industry sentiment that interoperable standards could reduce friction.

The UN's Independent International Scientific Panel on AI released a preliminary report in July identifying both opportunities and risks, including misinformation, discrimination, and cyberattacks. The Security Council meeting aims to translate these findings into political momentum.

Challenges Ahead: Verification and Enforcement

While the meeting may yield political signaling, a binding UN regime remains unlikely in the near term. OpenAI's stance—backing basic global safety standards without endorsing a specific international regulator—could be easier for governments and companies to accept than a new institution with enforcement powers. But it leaves unresolved who would independently verify compliance and what consequences would follow from non-compliance.

The inclusion of Chinese AI firms is a positive step, yet deep disagreements persist over verification, military uses of AI, export controls, and data governance. A global framework would need to bridge these gaps, a daunting task given geopolitical tensions.

"The key test after the speech will be whether industry commitments become independently verifiable and internationally interoperable," said Yoshua Bengio, co-chair of the UN AI panel, in an interview. "Without that, we risk a patchwork that fails to address catastrophic risks."

As the meeting unfolds, market participants will watch for any concrete commitments. OpenAI's recent executive turnover—including a new CRO and shifts in product leadership—adds another layer of complexity as the company balances commercialization, infrastructure build-out, and safety diplomacy.

Correction: An earlier version of this article misstated the date of the EU AI Act's general-purpose AI obligations. They began applying in August 2025, not August 2024.