Fetching from the wire…
01
02
03
04
05
06
07
08
Models2026-08-14 · source-backed
The alignment team published an aggregate benchmark for argumentation on questions that can't be empirically settled: 60% LMCA (560 position texts, 1,461 expert-rated arguments), 20% ACCoRD (~14,000 constraints checking probability consistency), 20% DTBench (407 expert-written decision-theory questions). Scores have risen roughly linearly since late 2024 with no saturation, though the team expects LMCA to saturate within a year. Models still do worse on anything that can't be verified empirically or mathematically.
Each link below shares sources, entities, or timing with this story.
Anthropic released Claude / Shared entities / Shared topic / Earlier coverage / Tension
Linked by a graph relationship (Anthropic released Claude); both cover Anthropic, Opus; overlapping topics (against, anthropic).
Jack Clark works at Anthropic / Shared entities / Shared topic / Earlier coverage / Tension
Linked by a graph relationship (Jack Clark works at Anthropic); both cover Anthropic, Opus; overlapping topics (against, anthropic).
Anthropic released Mythos / Shared entities / Shared topic / Earlier coverage / Tension
Linked by a graph relationship (Anthropic released Mythos); both cover Anthropic, Opus; overlapping topics (against, anthropic).
Anthropic released Claude / Shared entities / Shared topic / Earlier coverage
Linked by a graph relationship (Anthropic released Claude); both cover Anthropic, Opus; overlapping topics (against, anthropic).
Linked by a graph relationship (Anthropic released Claude); both cover Anthropic, Opus; overlapping topics (against, anthropic).
Copilot CLI supports Anthropic / Shared entities / Shared topic / Earlier coverage
Linked by a graph relationship (Copilot CLI supports Anthropic); both cover Anthropic, Opus; overlapping topics (anthropic, benchmark).
Anthropic released Mythos / Shared entities / Shared topic / Earlier coverage
Linked by a graph relationship (Anthropic released Mythos); both cover Anthropic, Opus; overlapping topics (anthropic, benchmark).
Anthropic released Claude / Shared entities / Shared topic / Earlier coverage
Linked by a graph relationship (Anthropic released Claude); both cover Anthropic, Opus; overlapping topics (anthropic, benchmark).