Writing

Notes, essays & research logs

These are working notes from my transition into AI safety, including essays on the problems I find most pressing and logs from research in progress. Some of it will be wrong, and that is the point of writing it down.

2026-08

Did we remove the misalignment, or just move it?

A negative result from our model-poisoning inoculation project: the defense we were most excited about did not contain the bad behavior. Instead, it made the behavior reachable from a prompt we never trained on.
2026-06

Building a research career in public

I am moving from production ML into full-time AI safety research. I'm early in the field, and I'm doing the work and the thinking out in the open.
2026-05

Why does inverse scaling happen? A research log

More test-time reasoning sometimes makes models worse. The literature documents that this happens; I want to understand, mechanistically, what goes wrong inside a single reasoning trace.
2026-04

Open-weight safety as a threat model

Safety properties that hold for a model in the lab rarely survive fine-tuning, quantization, or composition into real deployments. Open-weight release is where that gap becomes concrete and unrecoverable.

Notes, essays & research logs

Did we remove the misalignment, or just move it?

Building a research career in public

Why does inverse scaling happen? A research log

Open-weight safety as a threat model