max_paperclips

X · 54 atoms · 2026-06-24 → 2026-06-29 · 信用档先验 cred3

🟦 fact 4 · 🟥 take 50 · stance ▲22/▼14/◆10

覆盖实体 (top 10)

• OpenAI — 5 atoms
• Anthropic — 4 atoms
• DeepSeek — 3 atoms
• Mythos — 2 atoms
• state contracts — 2 atoms
• Dario — 2 atoms
• hosted agents — 2 atoms
• small models — 2 atoms
• NousResearch — 1 atoms
• US neolabs — 1 atoms

话题分布 (top 8)

政策管制 (2) · 模型禁令 (2) · 蒸馏质疑 (2) · 管理层可信度 (2) · 产品改进 (1) · 创新活力 (1) · 开源策略 (1) · 技术整合 (1)

样例 atoms

• 2026-06-26 · DeepSeek · Deepseek allegedly did 150k "interactions", literally nothing
• 2026-06-25 · GLM · GLM doing GOOD long horizon PPO
• 2026-06-25 · Chinese AI labs · Chinese labs are perfectly capable of paying people to assemble and annotate datasets (better than the $5 a day Nigerian sweatshops too most likely). They're producing world models
• 2026-06-25 · regulatory-driven AI developme · now it'll be explicitly and soley for weaponisation. maximum s-risk now
• 2026-06-28 · hosted agents · hosted agents are going to be huge, once the whole UX is clean

← Author 索引 · 实体目录