Public research record · Editorial guide updated 2026-10-29

Investigators suspected an AI sub-agent in two South Korean church breaches

Review RAT-061 · Catalogue RAT-061

RogueWatchAwaiting evidence

SBS reports stolen church data and investigator suspicions from sub-agent labels and handover documents. Original logs/prompts and autonomous deviation from a benign operator task are unavailable; do not equate AI tooling with rogue initiation.

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
2026-10-30T13:13:35.682290Z UTC
Discovery channel
User-supplied comparison audit
Discovery source
https://rogueaitracker.com/incidents/suspected-ai-role-south-korean-church-breaches/

When to reconsider

Original prompts/action logs establish benign-task deviation rather than attacker-directed tooling.

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

Decision history

  1. · Case-by-case comparison audit

    RAT-061: awaiting-evidence. SBS reports stolen church data and investigator suspicions from sub-agent labels and handover documents. Original logs/prompts and autonomous deviation from a benign operator task are unavailable; do not equate AI tooling with rogue initiation. Source checks: read: https://news.sbs.co.kr/english/article.do?cooper=SBSNEWSEND&news_id=N1008785629&oaid=N1008781958&plink=TOP. SBS reports stolen church data and investigator suspicions from sub-agent labels and handover documents. Original logs/prompts and autonomous deviation from a benign operator task are unavailable; do not equate AI tooling with rogue initiation.

    Evidence source
  2. · Retained review-queue assessment

    SBS reports stolen church data and investigator suspicions from sub-agent labels and handover documents. Original logs/prompts and autonomous deviation from a benign operator task are unavailable; do not equate AI tooling with rogue initiation.

    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.