Ahead of Sam Altman's UN address, OpenAI urges the US to lead an effort to develop global safety and security standards for building frontier systems
Maria Curi /Axios:
Context & Ripple Effects
OpenAI has framed AI safety as an international coordination problem since Altman called for China to help shape guidelines in 2023. By 2026, that position had narrowed toward a US-led architecture: a proposed international forum for AI standards would assess capabilities and risks while making models available to US allies.
The proposal also extends OpenAI's domestic policy agenda. Its June paper sought mandatory CAISI-led cyber-risk evaluations for advanced systems, rather than relying on the NSA; the UN-facing push takes the same safety-and-security framing into standard-setting diplomacy.
First-order effects
- US officials and UN delegates gain a concrete OpenAI-backed agenda for common technical safety and security standards for frontier systems, with US leadership as the organizing principle.
- OpenAI positions its own safety proposals alongside government-led oversight, making technical evaluations and security practices central to the company's policy case.
Second-order effects
- Frontier-model developers seeking access to US-aligned markets face greater pressure to participate in interoperable safety and security practices if a US-led standards forum takes shape.
- The earlier proposal to pair standards work with model availability for allies would make compliance and access part of the same diplomatic bargain, increasing the strategic value of shared technical rules.
Third-order effects
- If governments adopt the US-led approach, frontier-AI governance may be shaped less by standalone national rules and more by coalitions that combine security standards, market access, and alliance policy.
- The contrast with Altman's 2023 call for China to help shape guidelines signals a broader contest over whether international AI norms are negotiated universally or built through aligned blocs.
The trend: Frontier-AI governance is moving toward state-backed technical standards in which safety coordination and geopolitical alignment are increasingly intertwined.