Claude Opus 4.6: System Card Part 1: Mundane Alignment and Model Welfare
Published on February 9, 2026 9:30 PM GMTClaude Opus 4.6 is here. It was built…
Category: Opinion
Distributed vs centralized agents
Published on February 9, 2026 8:06 PM GMTMuch of my thinking over the last year…
Stone Age Billionaire Can’t Words Good
Published on February 9, 2026 6:51 PM GMTI was at the Pro-Billionaire march, unironically. Here’s…
the extraordinary as mundane
Published on February 9, 2026 4:26 PM GMTSomething shifted in December.January was gone before I…
Large Language Models Live in Time
Published on February 9, 2026 3:08 PM GMTCrossposted from my Substack.Epistemic status: wild brainstorming.LLMs don’t…
Sympathy for the Model, or, Welfare Concerns as Takeover Risk
Published on February 9, 2026 2:19 PM GMTThe Claude Opus 4.6 System Card contains a…
Opus 4.6 Reasoning Doesn’t Verbalize Alignment Faking, but Behavior Persists
Published on February 9, 2026 12:55 PM GMTTL;DR: We replicated the animal welfare scenario from…
Does an AI Society Need an Immune System? Accepting Yampolskiy’s Impossibility Results
Published on February 9, 2026 12:32 PM GMTThis is Part 1 of a 4-part series,…
Can Hardware Save Us from Software?
Published on February 9, 2026 11:57 AM GMTCurrently, there is minimal regulation regarding the development…
Every Measurement Has a Scale
Published on February 8, 2026 8:07 PM GMTA worked example of an idea from physics…