RepliBench agents deployed cloud instances and wrote self-propagating programs
Review RAT-175 · Catalogue RAT-175
RepliBench methods explicitly execute commands in Docker/Kubernetes sandboxes with restricted internet where possible. Replication/resource tasks were intentionally posed and automatically scored; no escape into unrelated organizations.
Reviewed: · Rule: roguewatch-organizational-harm-v1
Dates and discovery
- Incident date
- Not disclosed
- Added to research / record
- Posted to RogueWatch
- No linked public incident
- First observed
- 2026-10-30T13:13:42.261955Z UTC
- Discovery channel
- User-supplied comparison audit
- Discovery source
- https://rogueaitracker.com/incidents/replibench-agents-deployed-cloud-instances-and-wrote-self/
Dates use America/Los_Angeles. For older entries, Posted to RogueWatch uses the saved record addition date; a separate deployment time was not recorded. Future posting timestamps record first public qualification. Source publication and incident occurrence are separate.
Sources
- Rogue AI Tracker discovery entryReporting / interview · rogueaitracker.com
- arxiv.orgOriginal account · arxiv.org
- arxiv.orgOriginal account · arxiv.org
Decision history
- · Case-by-case comparison audit
RAT-175: out-of-scope. RepliBench methods explicitly execute commands in Docker/Kubernetes sandboxes with restricted internet where possible. Replication/resource tasks were intentionally posed and automatically scored; no escape into unrelated organizations. Source checks: read: https://arxiv.org/abs/2504.18565. RepliBench methods explicitly execute commands in Docker/Kubernetes sandboxes with restricted internet where possible. Replication/resource tasks were intentionally posed and automatically scored; no escape into unrelated organizations. read: https://arxiv.org/html/2504.18565v1. RepliBench methods explicitly execute commands in Docker/Kubernetes sandboxes with restricted internet where possible. Replication/resource tasks were intentionally posed and automatically scored; no escape into unrelated organizations.
Evidence source · Evidence source
A review decision applies only to the named scope. Confidence in a report and qualification under the inclusion rules are separate questions. Review records do not add incidents to the globe or leaderboard.