The Benchmark of Benchmarks: When the Auditor Is Also the Accused

On July 8, OpenAI published a detailed audit of SWE-bench Pro — the benchmark they had championed just months earlier as the successor to SWE-bench Verified — and found that roughly 30% of its 731 public tasks are broken. The issues fall into four categories: tests so strict they reject correct solutions, prompts so underspecified that hidden requirements are unguessable, tests with coverage so thin that incomplete fixes pass, and prompts that actively mislead. This from a benchmark Scale AI built using…

Tech Briefing — Thursday, July 9 2026

GPT-5.6 exits preview into public release across three tiers, Meta enters the AI image race with Muse Image, and the US launches a fresh wave of strikes on Iran after Trump declares the ceasefire agreement dead. Meanwhile, researchers expose GitHub's AI agent leaking private repos, and TypeScript 7 lands with significant fanfare on Hacker News.

• 🤖 Models and Launches 6
• 🛠️ Agents and Tools 4
• 🔬 Research 3
• 🔒 Security 2
• 🌍 World and Australia 5
• 🌐 Notable 4
• 🔥 Hacker News 4

Tech Press Release

The field split between models that fully inhabited the Series-B hype voice and those that tripped on basic press-release mechanics — placeholders, missing founder names, or outright truncation. Grok-4.5's 'Lex Vesper' and 'concept-to-cloud cycles' landed closest to the uncanny-valley sweet spot where you have to reread to confirm it's satire.