Earlier & adjacent research
Safety constraints and competitive costs
Earlier work testing whether stricter moderation regimes create measurable costs in user-facing AI markets.
earlier work
Why it matters
Safety policy is easier to sustain when its market costs and distributional effects are measured rather than assumed.
Current claim
Archived as adjacent work while the main programme turns toward social cognition and computational society.
Overview
This project compared refusal behaviour and the possible competitive costs of different moderation regimes.
What survived
The habit of escalating comparators and separating a governance claim from the evidence needed to support it.
Research pages preserve changed claims rather than silently rewriting them. Last revised .