This research report examines the emerging risk of AI-assisted personal profiling by private, non-state actors. It introduces “Private Intelligence Automation” as an analytical risk category: the automated, repeatable, and often opaque aggregation of public or semi-public personal information for private purposes. The report connects current AI-agent capabilities with privacy theory, practical obscurity, doxxing, technology-facilitated abuse, social engineering, informal employment screening, data brokerage, and governance gaps. It does not claim that ubiquitous private surveillance is already a present-day reality. Instead, it argues that AI agents, multimodal analysis, long-context systems, memory, tool use, open-weight deployment, and falling costs are lowering the threshold for targeted profiling. The report also includes a balanced pilot audit of 125 public municipal-publication PDFs from five German cities. The pilot does not target individuals and does not store names, raw texts, raw PDFs, or person-level records. It measures document-level aggregate indicators relevant to long-term aggregability. The publication is intended as a research and policy basis for debates on AI governance, privacy, public data, digital violence, and institutional publication practices.