Values

A short manifesto, uncited. Every other page on this site is sourced; this page is the editorial position.

AI safety is unsolved

Frontier AI is being deployed faster than the field can verify it is safe. The plausible failure modes include existential risk. Vendors publish responsibility pages, but cannot characterize what their own models will do.

Dangerous Robot treats every vendor safety claim as a claim, not a fact.

AI is fueling environmental destruction

Large AI models burn electricity, water, and embodied carbon at industrial scale. Firms are rushing to build fossil fuel-powered data centers to win an arms race. The costs are real and worth naming.

Our contradictions

Dangerous Robot runs on AI to investigate AI. We critique tools we depend on and burn electricity doing it. The contradictions are real, and we won't dress them up.

We try to be careful: small decisions go to small models, frontier models only when the task demands it. Our pipeline runs on Infomaniak, a Swiss provider using 100% renewable energy that recovers waste heat to warm buildings. It's a small effort, not a solution.