White House asks OpenAI to limit its next model release
The White House has requested that OpenAI restrict the release of its next AI model, citing national security and safety concerns.
Background
- OpenAI is the creator of ChatGPT and the GPT series of AI models (GPT-4, GPT-4o, etc.). Its "next model release" would be a major step forward in capability.
- The White House asking a private AI company to limit a release is highly unusual — there is no legal mechanism for the US government to unilaterally block open-source or commercial AI model releases (unlike, say, export controls on chips).
- This reflects a growing tension between AI safety/risk concerns (catastrophic misuse, loss of control, economic disruption) and industry pressure to ship products fast.
- Prior context: In late 2023, OpenAI reportedly paused or slowed GPT-5 training amid a leadership crisis and safety debates. The Biden administration also issued a 2023 executive order on AI safety, though its enforcement power is limited.
- The core unresolved question: who decides "how much capability is too much" in AI — the company, the government, or some third party?