Huo Xing Finance reports that last week, OpenAI CEO Sam Altman presented employees with three approaches to safely develop superintelligent AI. The first is to establish a global regulatory body, similar to the International Atomic Energy Agency, to inspect advanced models and audit safety measures. Insiders say Altman favors this path. The second approach emulates CERN, whereby major AI labs merge into a single organization or collaborate through a unified entity to jointly develop AI, rather than continuing to race independently. The third is the most extreme, dubbed by him the "Eliezer方案": one AI company achieves recursive self-improvement first, then leverages overwhelming economic power to control computing resources, preventing other developers from catching up. Altman described this idea as "melting down all other companies' GPUs," while emphasizing that the proposal is extremely problematic and intended only to fully outline possibilities. The "melting GPUs" concept originates from AI safety researcher Eliezer Yudkowsky, who proposed it in 2021 as a thought experiment—suggesting a superintelligent AI could create self-replicating nanomachines to melt global GPUs—to illustrate the extreme scale of intervention required, not as a recommendation for implementation. OpenAI previously proposed creating an "IAEA for AI" in 2023, and Google DeepMind’s CEO Demis Hassabis has also advocated for an "AI version of CERN." Currently, OpenAI, Google, and Anthropic plan to establish a joint safety standards organization by the end of this year or early next year, though they remain divided on responsibility for model safety testing. Altman seeks to impose speed limits on AI development but acknowledges that OpenAI cannot currently relinquish its competitive advantage to rivals.
OpenAI CEO Proposes Three Paths for Safe Super AI Development
MarsBitShare
Summary
Last week, OpenAI CEO Sam Altman outlined three strategies for safely developing superintelligent AI to his staff. One option is a global regulatory body, similar to the IAEA, to audit safety measures and align with CFT protocols. A second proposal suggests merging leading AI labs into a CERN-like organization. The third, the “Eliezer Plan,” would involve a single company achieving self-improvement and dominating computing power—a scenario Altman dismissed. OpenAI, Google, and Anthropic aim to launch a safety standards group by late 2026, but disagreements remain over model testing, including protections for safe-haven assets.
Source:Show original
Disclaimer: The information on this page may have been obtained from third parties and does not necessarily reflect the views or opinions of KuCoin. This content is provided for general informational purposes only, without any representation or warranty of any kind, nor shall it be construed as financial or investment advice. KuCoin shall not be liable for any errors or omissions, or for any outcomes resulting from the use of this information.
Investments in digital assets can be risky. Please carefully evaluate the risks of a product and your risk tolerance based on your own financial circumstances. For more information, please refer to our Terms of Use and Risk Disclosure.