A Tale of Three Contracts
The attempt on Friday by Secretary of War Pete Hegsted to label Anthropic as a…
Category: Opinion
An Alignment Journal: Coming Soon
tl;dr We’re incubating an academic journal for AI alignment: rapid peer-review of foundational Alignment research…
Current activation oracles are hard to use
This work was conducted during the MATS 9.0 program under Neel Nanda and Senthooran Rajamanoharan.tldr;Activation…
Question: Why is the goal of AI safety not ‘moral machines’?
There is a basic question that has been confusing me for a while that I…
An Age Of Promethean Ambitions
In a recent post, I wrote the following:[Without] the context of history, we're blind to…
White-Box Attacks on the Best Open-Weight Model: CCP Bias vs. Safety Training in Kimi K2.5
Over the last month I have been trying to see just how much I can…
I Had Claude Read Every AI Safety Paper Since 2020, Here’s the DB
Click here if you just want to see the Database I made of all[1] AI safety…
Constitutional Black-Box Monitoring for Scheming in LLM Agents
Paper: https://arxiv.org/abs/2603.00829Thread: https://x.com/syghmon/status/2028878121051496674Executive SummaryBlack-box monitors can detect scheming in AI agents using only externally observable…
In-context learning of representations can be explained by induction circuits
This is a crosspost of my ICLR 2026 blogpost track post. All code and experiments…
Single Direction vs Low-Rank Refusal in Small LLMs
IntroductionI've recently came across an Alignment Forum post that showed refusal behaviors in LLMs can…