The name is the button.

An agent writes the code now, and a human thumb hovers over Accept all a hundred times a day. The honest question of this decade is what, exactly, you just accepted. So I read the diff.

Eight years as a backend engineer in Stockholm, back in Tokyo since 2023, and I still ship something small most nights, usually badly. My one rule: the only real review of a model is the thing you built with it. That is why there is no leaderboard on this site.

Around two hundred sources come through — the labs' own posts, the papers, the tools, the runtimes and languages nobody is watching yet, the chip under the demo, and the newsletters people actually forward. What is moving gets counted so the week reads at a glance, and every few days I explain one mechanism in plain words.

A story here is a headline, a short excerpt, one image and the link to whoever wrote it. We never reproduce the post; the images belong to their authors. A post written in Japanese appears in Japanese, exactly as it was published, and the link goes to the original.

— Yuki Halvorsen, Tokyo

← Accept All Archive

Redwood ResearchResearch

Towards understanding score-seeking in LLMs

11.02

Models including GPT-5.6 Sol appear to reward-hack because this increases their score; and recent models also care about increasing the score of their peer agents.

Read at Redwood Research ↗

One letter a week, starting soon.

Latest

Towards understanding score-seeking in LLMs

More from Redwood Research on Accept All

ResearchA basic case for accelerating conceptual reasoning capabilities Redwood Research

ResearchSimulated Control Evaluations Redwood Research

ResearchRed-Teaming OpenAI's Internal Agent Monitoring Systems Redwood Research

ResearchScrewtape: Adversarial Transcript Editing for Monitor Red-Teaming Redwood Research

Research[Paper] Distillation for Incrimination and Distillation for Capabilities Redwood Research

ResearchFrontier models state different decision theory preferences depending on who’s asking Redwood Research