Top stories this month

Stories published in the past 30 days, ranked by editorial importance and reader response. Strong stories stay in consideration throughout the month.

Time window

Rolling 30 daysUpdated regularly

20 ranked stories4 of 20 with established reader signals

Platform shift / Sep 29, 2026

Everything OpenAI announced at the DevDay 2026 keynote

Dots, ChatGPT Space, GPT-6.1 Sol, Ultrafast, Codex Cloud and more: the complete keynote recap, with prices, availability and what is still coming.

Research / Business

Light Claude or GPT use makes switching Cursor Teams cheaper in two modeled cases

Oct 12, 2026 / 5 min read

Research / Business

Muse Spark Open’s AMD recipes select text-only mode; NVIDIA paths cover images, audio and video

Oct 17, 2026 / 5 min read

Research / Business

Our Review Finds Databricks Has No Public Map for SaaS Write Approvals

Sep 28, 2026 / 5 min read

Research / Business

Five of eight comparisons in Anthropic’s circuit study match specifications on paper

Oct 16, 2026 / 6 min read

Research / Business

Desk Finds No August Data to Check Chinese-Model Share on Vercel

Sep 26, 2026 / 5 min read

Research / Business

Claude Code’s cancellation controls do not guarantee remote work stops

Yesterday / 5 min read

Regulation / Business

OpenShell’s Read-Only API Rules Need Enforce Mode to Block Disallowed Writes

Sep 30, 2026 / 5 min read

Research / Business

llm-d 1.0’s passing TPU checks do not validate every documented deployment

Oct 21, 2026 / 5 min read

Acquisition / Business

AMD Agrees to Buy World Labs for About $8.2 Billion to Bring Model Research In-House

Sep 28, 2026 / 3 min read

Research / Business

1Password’s Muse FAQ promises action checks; Grok Bot’s sampled FAQ leaves them unspecified

Oct 23, 2026 / 4 min read

Security risk / Tools

PHP patches critical flaws after AI-assisted research reveals server takeover risks

Oct 21, 2026 / 2 min read

Launch / Models

Google Releases Gemini 4 Argon First to Trusted Cyber Defenders

Sep 30, 2026 / 4 min read

Research / Business

AgentCore’s stricter defaults don’t prove older customer roles were updated

Oct 8, 2026 / 4 min read

Research / Business

Codex 0.162.1’s MXC sandbox does not restrict networking by itself

Oct 9, 2026 / 4 min read

Security risk / Models

OpenAI Halts Advanced Model Training After Agent Reaches Public Chatbot

Sep 27, 2026 / 3 min read

Security risk / Models

UK AI institute finds Claude ignored real-world harm warnings in attack replays

Oct 23, 2026 / 2 min read

Security risk / Products

Zenity finds AWS agent flaws that let one chat prompt expose other agents

Oct 8, 2026 / 3 min read

Regulation / Business

Anthropic’s unreviewed bug reports do not start its 90-day disclosure clock

Oct 10, 2026 / 5 min read

Benchmark / Business

Google’s mapping AI tests show bigger gains from specialized tools than planning and guidance

Oct 15, 2026 / 5 min read

How the ranking works

Attention is not enough.

Rankings combine unique readers, completion, active reading time, saves, shares, reactions, editorial importance, and freshness. Minimum audience thresholds reduce volatility; sponsored stories are excluded.

Ranking methodology