I consolidated Decision-OS V4–V8 into a single gateway (Zenodo DOIs + SSOT). Key idea in V8: evaluate/control *trajectories* (Time-Tube), not point outputs. Self-Safe = harm / collapse / dependency gradient (+ reversibility interventions) AGI論文を書きました。良かったら読んでください。 Gateway: https://github.com/shin4141/decision-os-paper/blob/main/README.md
- 4 replies
- 4 recasts
- 12 reactions
AI safety is not only about model behavior. It is also about what happens before an irreversible action executes. Decision-OS V5 Revised (SiriusA2) proposes a confirmation layer for human–AI decision safety: intent → observe/hold → approve/reject → execute/revoke It combines: * trajectory-aware duress routing * two-step confirmation * revoke path * family multisig * ZK authorization without exposing PII * non-PII audit evidence ZK does not read the user. It proves authorization. That distinction matters. Preprint: <https://zenodo.org/records/19828435>
- 0 replies
- 1 recast
- 4 reactions
30+ AI debates. Now in English. Most people pick a side. Few can see where it breaks. Try it:https://mmar-l0-core.onrender.com/gallery?lang=en Drop a tweet. I’ll turn it into an AI debate.
- 1 reply
- 4 recasts
- 8 reactions