Back

06 / SafetyOS · AI safety

SafetyOS / AI safety

Mapping AI risk in public.

SafetyOS is an open platform for reducing existential and systemic AI risk through structured global discourse, live risk monitoring and expert-verified debate.

RoleCreator / Engineer
ScopeProduct · Forum systems · Web

Risk argued in the open.

  1. 01Structured discourse

    Open participation only compounds when the structure holds it together.

  2. 02Evidence attached

    Claims, sources and uncertainty stay visibly connected.

  3. 03Legible, not authoritative

    Public-interest software earns trust by showing its working.

02 / System card

A compact record of what the project made, taught and runs on.

Project system / 06

SafetyOS

Work in progress

Where it got to

End-to-endplatform, built solo

LiveAI risk map

Openpublic discourse model

What we learned

  1. AI safety needs a public square, not only a research literature. Most of the argument still happens in closed rooms.
  2. Structure is what turns open participation into knowledge instead of noise.
  3. Trust is earned by showing the working: claims, sources and uncertainty have to stay attached to one another.

Tech stack

  • Next.js
  • React
  • TypeScript
  • PostgreSQL
  • Authentication
  • Shared UI

Project state

Unlaunched and still being built. I care about AI safety and wanted somewhere the argument could happen in public, so this is an attempt to start that conversation rather than a finished product.

Novica Petrovic / Selected workSystem record · SafetyOS
NextSpecframe