Miles_Brundage

X · 27 atoms · 2026-06-23 → 2026-06-28 · 信用档先验 cred3

🟦 fact 2 · 🟥 take 25 · stance ▲11/▼7/◆7

覆盖实体 (top 10)

• GLM 5.2 — 4 atoms
• OpenAI — 3 atoms
• Anthropic — 2 atoms
• Google — 2 atoms
• 美国政府 — 2 atoms
• xAI — 1 atoms
• AI safety accidents — 1 atoms
• Obernolte-Trahan bill — 1 atoms
• Biden admin — 1 atoms
• US government — 1 atoms

话题分布 (top 8)

安全领导力 (3) · 监管框架 (2) · 内部安全管控 (1) · 国籍限制 (1) · 产业角色转变 (1) · 合作伙伴关系 (1) · 安全风险 (1) · 自动化偏差 (1)

样例 atoms

• 2026-06-25 · Biden admin · Not sure what you mean by the Biden admin 'emphasiz[ing] safety provisions such as mandatory model review' but there was nothing like this under the Biden admin
• 2026-06-23 · xAI · When xAI rents a datacenter to Anthropic, it goes from being a country of ingenues in a datacenter to a country of geniuses in a datacenter
• 2026-06-24 · AI safety accidents · Automation bias is a common, well-documented factor in many safety accidents (aerospace, nuclear power, etc.). The difference with AI is that ppl racing towards accidents loudly tw
• 2026-06-24 · GLM 5.2 · also are forced by necessity to eke out all the efficiency gains they can
• 2026-06-24 · GLM 5.2 · scaling laws generally are sublinear, ie you can get much more than 1/10th the performance with 1/10th the compute

← Author 索引 · 实体目录