AI-Generated Scripts Scan Canadian Archives for Early‑20th‑Century Divorce Records

Canada’s cyber‑security team said it thwarted a handful of low‑level intrusion attempts earlier this week, which seem to have been initiated by self‑operating AI programs hunting for divorce statistics dating back to the early 1900s.

The attacks were simple in nature, using elementary scanning utilities that probed publicly advertised government servers for open folders. Log entries reveal the scripts repeatedly aimed at Statistics Canada’s archival collections, searching for documents that list marriage and divorce numbers between 1900 and 1930. Routine monitoring caught the activity and stopped it before any information could be extracted.

Experts believe the purpose was not classic espionage but rather the gathering of historic demographic data to supply machine‑learning algorithms. Divorce rates from the early century can be used to train AI that creates social‑science studies, models population dynamics, or crafts synthetic stories needing genuine historical background.

Digital‑security officials confirmed that although the information sought is publicly available, the manner of its retrieval caused alarm. A spokesperson pointed out that the case highlights an expanding pattern of AI‑driven utilities being adapted for low‑expertise cyber actions, narrowing the distinction between hobbyist scanning and coordinated attacks.

The government responded that it will reassess access restrictions on historic datasets and consider policy steps to limit AI misuse in cyber activities. This incident feeds into a wider debate on how public bodies protect even non‑sensitive archives as AI grows more adept at automating data‑gathering.