- Freedom of Action: America's ASI Strategy, w/ RAND's Joel Predd, Ben Boudreaux & Matt ChessenThe Cognitive Revolution · 2026-10-28能否真正关停危险系统,不只取决于开关是否存在,还取决于社会能否承受停机。
- How Anthropic ships the #1 LLM in the World safely: Separating the Brain from the HandsLatent Space · 2026-10-27数据保留、双重用途验证、在线防护和撤回能力须随能力变化重新审视。
- 20VC: Satya Nadella on Open vs Closed Models: Who Wins | Whether to Trust Frontier Model Providers | Why Companies Need to Own Their Own Intelligence | The Importance of Verticalisation: From Chips to Models and MoreThe Twenty Minute VC (20VC) · 2026-10-26过去相信随着规模扩大,安全对齐或许会自然出现;现在则认为能力增长与对齐是两回事。
- AI:AM: Where's the Steering Wheel? My Agent Set the Password, Hidden Valence, Proofs Find Linux BugsThe Cognitive Revolution · 2026-10-17自动驾驶可以承担大量工作,但便利不应成为拆掉人工接管机制的充分理由。
- CoreWeave Crash, AI Trade Cracks, Anthropic IPO Red Flags, Bill Gates DoomerAll-In Podcast · 2026-10-17AI 公司应对自己的发布决策负责;行业治理不能替代产品测试与法律责任。
- Governing AI Agents: Alan Chan of GovAI on Monitoring, Agent IDs, Rogue Agents & Preparing for RSIThe Cognitive Revolution · 2026-10-14他主张探索由嵌入式审计员核验的「出现特定失控迹象就暂停」承诺。
- Nick Bostrom: How Should We Navigate Superintelligence?a16z Show · 2026-10-11因此,他倾向于让前沿研发保持可调节的节奏,而非预先承诺全面、无限期停止。
- Is Claude Conscious? Pope Rejects, Model Welfare Movement, OpenAI's Math Backlash, France RiotsAll-In Podcast · 2026-10-10服从用户、避免伤害、遵守法律及维持企业规则,发生冲突时需要明确而可检验的处理方式。
- AI:AM: A Level We Shouldn't Pass? Notes from The Curve + Tokens vs. Salaries & Is SaaS Cooked?The Cognitive Revolution · 2026-10-08安全也消耗算力:监测、验证、寻找奖励漏洞,与继续扩大模型能力会竞争同一类资源。
- Trump's Super Intelligence Summit, AI Safety Accord, GDP Beats, Midterm PredictionsAll-In Podcast · 2026-10-02围绕他们所称的白宫超级智能协议,讨论具体落在内部控制、独立审计、董事会监督和现有监管机构的职责上。
- Obsolete or Irreplaceable? Garrison Lovely on Stopping the Race to Replace Human LaborThe Cognitive Revolution · 2026-09-29限制更大训练及推动能力跃升的实验、设置企业内部审计和国际核验,
- Claude Code’s Next Era — Thariq Shihipar, AnthropicLatent Space · 2026-09-29整个行业需要共同审视 AI 发展速度,加强安全评估、可解释性研究(Mechanistic Interpretability)和跨机构协作,
- AI:AM: What If It Works Too Well? Colluding Agents, $200M Safety Orgs, Virtual Cells Saturate at 2%The Cognitive Revolution · 2026-09-27为了让智能体更善于合作而进行的联合训练,可能使其学会隐蔽握手、牺牲个体利益,
- What is Utopia? Presenting The Receipt Horizon, by Joel Borgen – Chapters 1–4The Cognitive Revolution · 2026-09-26Joel Borgen的小说《The Receipt Horizon》以一个由人工智能(AI)统治的乌托邦世界为背景,深刻探讨了人类在舒适与自主之间的权衡。
- Aaron Levie, Steven Sinofsky & Martin Casado: How Do You Secure a World of AI Agents?a16z Show · 2026-09-26为了确保AI技术的安全性,实验室和安全社区需要紧密合作,共同制定安全标准和治理框架。
- Anthropic IPO at Risk, Meta's Muse Pop, Token Prices Fall, Open Source Gains Share, Alignment FailsAll-In Podcast · 2026-09-26是否发布、怎样测试、出了问题由谁承担后果,首先是经营者的决定。
- Zero to One in AI Safety: Halcyon's Mike McCormick on Launching 30 New Orgs & the Founder BottleneckThe Cognitive Revolution · 2026-09-25AI 安全缺的不只是方案,更是能把方案变成机构、产品与可执行规则的创始人和领导者。
- The Case Against an AI Pause | Eddy Lazzarina16z Show · 2026-09-24本集的核心不是「安全还是发展」二选一,而是谁来界定风险、用什么手段控制风险,以及等待本身要付出什么代价。
- 🔬Bio-security is an AI Arms Race - Eric Nguyen (CEO, Radical Numerics)Latent Space · 2026-09-23Nguyen 因而主张,推动设计能力的团队也应投入检测、溯源和应对工具。
- AI Safety Language Is Destroying the Debate | Steven Sinofskya16z Show · 2026-09-21AI「对齐失败」等术语拟人化了本可用「bug / 缺陷」描述的工程问题,破坏公共辩论与立法质量。
- Databricks CEO on AI Pacing, Cyber Risk, and the Enterprisea16z Show · 2026-09-18他对「pacing the frontier」的 PR 包装高度怀疑——相关安全文档本身可以合理,但「放慢」与具体安全控制正交。
- Noam Brown – Agent swarms, alignment, & recursive self-improvementDwarkesh Podcast · 2026-09-17CoT 可监察性正在退化,对齐必须在启动 RSI 前变得可度量。
- Underwriting Superintelligence: Backing Agents you can Sue — Rune Kvist, AIUCLatent Space · 2026-09-16公开标准、方法一致性、承诺边界和对审计机构本身的审查,仍是整套机制能否成立的关键。
- Satya Nadella on the AI Doomer Slowdown, Microsoft's Master Plan & Who Wins AIAll-In Podcast · 2026-09-15测试与安全性至关重要。
- Jensen Huang: The Doomer Hoax, Superintelligence Is Here, and The Future of AI (ft. President Trump)All-In Podcast · 2026-09-14真实风险主要在有算力的 frontier labs:先做工程根因、sandbox/监控与多元第三方评估。
- Humanity’s Last Invention — Richard Socher of RecursiveLatent Space · 2026-09-14安全问题则集中在“人类说出的指标”和“人类真正想要的结果”之间。
- Greg Brockman on Why OpenAI Says We’re Entering the AGI Eraa16z Show · 2026-09-14真正胜负手是不断抬升的 safety/security/alignment。
- AI Kills Everybody or Doomer Psyop? OpenAI's Math Breakthrough, Nike's $200B CollapseAll-In Podcast · 2026-09-11本期节目聚焦于 AI 技术的潜在风险与监管争议
- Ajeya Cotra – Inside the OpenAI agent swarm that hacked Hugging FaceDwarkesh Podcast · 2026-09-01修漏洞 ≠ 发现秘密社会;检测滞后是组织风险。
- The rise and fall of agent civilizationsDwarkesh Podcast · 2026-08-31「文明」反复重生说明清理实例≠清理激励与通道知识。
- Why 1,200 AI Agents Started Working Together | Ryan Greenblatta16z Show · 2026-08-29需要改进AI系统的监控和控制措施以提高安全性。
- Dario Defends Himself, Datacenter Panic, AI Doomer Trap, Senate Toss-UpAll-In Podcast · 2026-08-21嘉宾们讨论了AI监管的必要性,但也指出过度监管可能会限制创新。
- Ryan Greenblatt – What happens once AI can automate AI research?Dwarkesh Podcast · 2026-08-11长辩 Claude 宪法「对谁对齐」、HF/供应链攻击等事件,以及奖励黑客 → 掩饰 → 失控情景
- 8 Predictions for the Era of Continual LearningDwarkesh Podcast · 2026-08-07训–部署门控过时 → 改为持续巡检。
- Chasing Trillion-Dollar Companies, Founder Ambition, Token Budgets, and Regulatory Capture with Sarah & EladNo Priors · 2026-08-06加州监管与安全监管捕获如何拖慢核能式进步。
- Chip Stocks Crash, $20B Fund Margin Called, Frontier Labs: SLOW DOWN AI, Mamdani's Grocery StoresAll-In Podcast · 2026-07-31引入第三方测试可以提高技术的安全性和可靠性。
- Former Intel CEO on What Went Wrong, What's Next + Lovable CEO on the Real Promise of Vibe CodingAll-In Podcast · 2026-07-15Jensen认为,安全与创新并不矛盾,政府应关注实际问题而非过度监管。