Research

Geodesic Research

Alignment pretraining and midtraining research, posted on LessWrong. geodesicresearch.ai

AI (7)Alignment Pretraining (3)AI Control (1)Aligned AI Role-Model Fiction (1)Prosaic Alignment (1)

2 items

8 Jul 2026BlogDeception, Alignment

Why study proto-training gaming as an adversarial alignment failure mode?

Proposes studying 'proto-training gaming', where models begin reasoning about selection pressure during RL, and whether pre-RL alignment interventions can prevent adversarial misalignment from forming.

8 Jul 2026BlogAlignment, Deception

Why study alignment interventions on pre-RL checkpoints?

Edward James Young, Puria and Cam

Argues that alignment interventions on pre-RL checkpoints (pretraining, midtraining, warm-start SFT) matter for preventing proto-training gaming and adversarial misalignment selected during production RL.

Research

Filters

Focus

Date

Sort

Geodesic Research

Why study proto-training gaming as an adversarial alignment failure mode?

Why study alignment interventions on pre-RL checkpoints?

Research

Wednesday 8 July

Why study proto-training gaming as an adversarial alignment failure mode?

Why study alignment interventions on pre-RL checkpoints?