Dario Amodei
CEO & Co-Founder, Anthropic
Co-founder & CEO of Anthropic. Former VP of Research at OpenAI. The most influential voice in AI safety. Predicts AGI by 2026–2027.
Anthropic 联合创始人兼 CEO，前 OpenAI 研究 VP。AI 安全领域最具影响力的声音，提出 Responsible Scaling Policy 框架。预测 AGI 在 2026–2027 年到来。
Framework · 理论框架
RSP 负责任扩展框架 / Responsible Scaling Policy
RSP 框架
AI Safety Levels 分级体系 / Tiered safety classification
→
ASL-2
当前水平，风险可控 / Current level
→
ASL-3
需额外安全措施 / Additional safeguards
→
ASL-4+
需全新安全范式 / New paradigm needed
Words · 金句
Notable Quotes
"我们正在建造比我们更聪明的东西，这需要极度的谨慎。 / We are building things smarter than us — that demands extraordinary caution."
"不要优化你不理解的东西。 / Don't optimize what you don't understand."
Appearances · 播客出场
Episodes
Anthropic Research · Alignment Series
Constitutional AI 2026: Training Claude to Self-Critique
Anthropic 详解 Constitutional AI 2026 版本的演进：从原则列表到 AI 自我对练的完整工作流。...
Anthropic Product Talks · Deep Dive
Computer Use: How Claude Learned to Operate Your Computer
Anthropic 工程团队详解 Computer Use 能力的技术原理、训练方法、安全机制和未来路线图。...
Dwarkesh Patel
Alignment, Anthropic's Research Culture, and the Road Ahead
Dario 与 Dwarkesh 深度对谈 AI 对齐的技术细节、Anthropic 独特的研发文化、以及在实际工程中实现安全承诺的方法论。...
Lex Fridman Podcast · #452
AGI in 2027, Scaling, and the Future of AI Safety
Anthropic CEO Dario Amodei 与 Lex Fridman 展开三小时深度对话，系统阐述了他对 AGI 时间线（2026-2027）的判断、Responsible Scaling Policy 安全框架的设计逻辑、以及...
Bloomberg Odd Lots
Doing Research at the AI Frontier
Anthropic 联合创始人兼顶级经济学家深度讨论 AI 前沿研究的商业逻辑、安全与 Scaling 的平衡。...
Anthropic Talks · Special
The Economics of AGI: Power, Compute, and National Strategy
Dario Amodei 在 Anthropic 主办的 AGI 经济论坛上系统论述 AGI 时代的算力集中、国家战略、权力分配与人类繁荣的辩证关系。...
Anthropic Engineering Blog · Deep Dive
Inside Anthropic's Compute Strategy
Dario 详解 Anthropic 在 AWS Trainium、Google TPU、NVIDIA GPU 之间的算力分配策略，及对自研芯片的思考。...
Anthropic Research Talks · Interpretability Series #3
Mechanistic Interpretability: How Anthropic Reads Model Minds
Anthropic 可解释性团队负责人详解如何使用 Sparse Autoencoders、Cross-Layer Transcoders 逆向工程 Claude 4 的内部电路。...