The AI Chernobyl Problem: Reciprocal Verification, Containment and Resilience in Autonomous Critical Systems
Consolidated working draft v0.1 of a proposed research programme on reciprocal verification, containment and resilience in autonomous critical systems. Proposed, not validated. Nothing has been run and no results exist anywhere in this programme. The first experiment is designed (draft pre-registration v1.2, not frozen) and has not been executed. Most of the programme's 17 deliverables are pending.
Authors
- Michael Pilaprat
Publication Details
- Journal
- Zenodo (CERN European Organization for Nuclear Research)
- Published
- 2026-11-07
- DOI
- https://doi.org/10.5281/zenodo.24046218
- Primary Topic
- Safety Systems Engineering in Autonomy
- Type
- preprint