ControlAI 执行董事康纳·利希:超级智能不是武器,而是对手
ControlAI’s Connor Leahy on why superintelligence is ‘not a weapon, it’s an adversary’
ControlAI 研究者兼企业家康纳·利希在 TechCrunch 节目中讨论超级智能的控制难题,以及 AI 安全事件带来的部署风险。
TechCrunch 报道称,ControlAI 的康纳·利希参加了 Equity 播客讨论。节目指出,AI 公司普遍把超级智能视为不可避免,但近期安全事件(包括 OpenAI 的 Hugging Face 入侵事件)显示,部署超越人类能力的系统可能带来难以控制的后果。利希将超级智能描述为“对手”,强调其风险不应仅按传统武器框架理解。
对 AI 行业的影响
这种框架会推动行业把 AI 风险管理从产品安全扩展到系统性对抗与失控预案,影响模型发布、监测和实验室间协调。它也可能强化对能力评估和部署限制的需求,但具体措施尚未由报道确认。
原文参考
来源:TechCrunch AI · 2026-09-09
AI companies have been talking about superintelligent AI like it’s inevitable, but recent safety incidents like OpenAI’s Hugging Face breach are demonstrating the potential dangers of deploying AI systems that are more capable than humans. So what happens when we can’t reliably control what these systems do? On this episode of TechCrunch’s Equity podcast, Rebecca Bellan is joined by Connor Leahy, an AI researcher, entrepreneur, and now the U.S. Executive Director of […]