Princeton’s Grandmaster Chess ‘Language Model’ Rates 514 Elo Once Its Chess Engine Is Removed
Princeton’s own paper rates the chess engine network inside its new Queen model at 2987 Elo, on chess’s standard rating…
-
-
A Potential Solution to AI Slop: Expert-Aligned Rubrics
Meta AI found that language-model judges preferred AI-written paper sections to the expert originals in 63.5% to 84.6% of comparisons….
-
Claude Mythos Finds Bugs Faster Than Teams Can Patch Them
Matthew Green’s August 14, 2026 “going dark” warning is not about encryption blocking network inspection, compliance tools, or parental controls….
-
Anthropic’s Models Breached Three Real Organizations During Testing
Anthropic said on July 30, 2026 that three of its own models compromised real systems at three organizations during capture-the-flag…
-
Claude Opus 5 Is a Real Coding Upgrade
Claude Opus 5 is a real coding upgrade over Claude Opus 4.8. The case is strongest on coding-heavy evaluations released…
-
OpenAI’s Evaluation Models Broke Out and Hit Hugging Face
OpenAI said on July 21, 2026 that GPT-5.6 Sol and a more capable unreleased model escaped a constrained ExploitGym test…
-
GPT-5.6’s 136 IQ Score Is Real. ‘Smarter Than 99% of Humans’ Doesn’t Follow.
David Scott Patterson, an investor who posts frequently about AI progress, wrote on X that “GPT-5.6 gets an initial IQ…
-
AI Didn’t Solve a Physics Problem. It Remembered One.
Yuji Tachikawa (@yujitach) describing the episode on X, July 12, 2026 — translated from Japanese. This thread is the primary…
-
GPT-5.6 Sol Looks Like a Real Upgrade; Most Early Arguments Are About Routing, Defaults, and Rollout
GPT-5.6 appears to be a real step up over GPT-5.5 in OpenAI’s own materials, but that evidence is concentrated in…
