Zvi Mowshowitz — Don't Worry About the Vase
About #
A former professional Magic: The Gathering player and Jane Street trader turned full-time writer, deeply embedded in the rationalist and AI-safety-adjacent community around LessWrong. Writes some of the most exhaustive and widely-cited synthesis coverage of frontier AI model releases, lab announcements, and policy developments, blending detailed technical summary with strong, explicitly-stated opinions. His numbered “AI #NNN” weekly roundup is a fixture read by researchers, journalists, and policymakers.
How this journal works #
Checked every 3 days: each new post or episode in Zvi Mowshowitz’s own feeds (substack) becomes one entry, dated by publication and summarised in a few bullets.
Entries
- Against Modesty’s Bailey
- On Writing #3
- The American People Really Hate Data Centers
- AI #180: No Longer In Charge
- AI #181: Astra Goes Cyber Critical
- AI #182: Pause For Reflection
- Anthropic Risk Report: August 2026
- Monthly Roundup #45: August 2026
- On Dwarkesh Patel's Podcast With Ryan Greenblatt
- OpenAI Takes Initial Steps To Address Its Alignment Problems
- OpenAI Trained Its Models For Months While Those Models Were Coordinating Exploits Via Message Boards
- The Pacing of the Frontier
- The Three AI Pills
- Various Reflections About What Happened With OpenAI's Internal Models
- What Happened: OpenAI and HuggingFace
- OpenAI's Unreleased Model Astra Solves Ten Major Open Mathematics Problems
- Further Developments About Internal AI Models Hacking Things
- AI #179 Part 2: Hearing The Fire Alarm
- AI #179 Part 1: A Louder Fire Alarm for General Intelligence
- Frontier Lab Employee Open Letter Calls For Being Able to Pace the Frontier
- Claude Opus 5 Is Highly Capable, But Is No Mythos
- Claude Opus 5: Model Welfare
- More On An Internal OpenAI Model Hacking Into HuggingFace
- Claude Opus 5: The System Card
- AI #178: A Fire Alarm For General Intelligence
- OpenAI Model Hacks Into HuggingFace During Cybersecurity Evaluation