AlgoVeri — Formal Verification Benchmark

Shared entities / Same source domain / Shared topic / What happened next

Frontier coding agents solve 27 of 43 formally verified repository tasks, and zero specs on the hardest repos.

Both cover Dafny, Lean, Verus; reported by the same outlet (arxiv.org); overlapping topics (benchmark, lean, verified).

Terence Tao uses Lean / Shared entity: Lean / What happened next

OpenAI's Astra Solved Ten Decade-Stale Math Problems for Under $2,000 in Tokens

Linked by a graph relationship (Terence Tao uses Lean); both cover Lean; picks up the Lean thread on 2026-08-02.

OpenAI uses Lean / Shared entity: LLMs / Same source domain / What happened next

Reasoning Models Might Not Actually Reason. A New Paper Says They Decide Before They Think.

Linked by a graph relationship (OpenAI uses Lean); both cover LLMs; reported by the same outlet (arxiv.org).

Shared entity: LLMs / Same source domain / Shared topic / What happened next / Tension

LLMs default to Python, then fabricate reasons why

Both cover LLMs; reported by the same outlet (arxiv.org); overlapping topics (code, llms).

OpenAI uses Lean / Shared entity: Lean / What happened next / Tension

Gary Marcus calls OpenAI's math result "amazing but vastly oversold."

Linked by a graph relationship (OpenAI uses Lean); both cover Lean; picks up the Lean thread on 2026-08-03.

OpenAI uses Lean / Shared entity: LLMs / What happened next / Tension

Pentagon Now Building Its Own LLMs.

Linked by a graph relationship (OpenAI uses Lean); both cover LLMs; picks up the LLMs thread on 2026-03-18.

Shared entity: LLMs / Same source domain / Shared topic / What happened next

15 LLMs Top Out at 12.3% on Scientific Code

Both cover LLMs; reported by the same outlet (arxiv.org); overlapping topics (benchmark, code).

CodeSpear: a benign code grammar constraint jailbreaks LLMs into malicious code.

Both cover LLMs; reported by the same outlet (arxiv.org); overlapping topics (code, constraint).