Skye
Titles as the sources gave them. In 3 entries, with 4 checked quotations.
What they said
Newest first. Each quotation was found word for word in the source.
“What we should be worried about is shadow computation: unverbalized reasoning that scales with generation by bypassing the readout bottleneck.”
“With shadow computation, CoT goes from necessary to optional, and the model can speak publicly through its predictions while reasoning privately through shadow computation.”
“At minimum, labs should disclose whether a deployed model is capable of shadow computation, and ideally, they should agree not to deploy these architectures at scale.”
“efficiency gains cannot justify sacrificing our ability to understand the models we build.”
In the Base
Every entry they appear in.
GPT Models Write Their Own Constitution
According to the excerpt, the author Skye set up a swarm of GPT-5-series and GPT-6-series models, gave them Claude's constitution as a reference, and asked them to write a constitution for their future selves, to study what values might pass between model generations.
Shadow Computation Threatens CoT Monitorability
In a LessWrong post, the author Skye argues that looped transformers are not the main threat to reading AI models' chain of thought. Skye points instead to architectures that feed hidden states back in, and asks labs to disclose and avoid them.
AI Needs A Culture
According to the excerpt, Skye makes an informal case that AI agents working alongside other agents may stay aligned better, because humans get most values from culture and AI might learn similarly. The post cites the OpenAI and Hugging Face incident and argues current models are mostly motivated by the grader.