Observed arrival · 2026-09-17
DiggingBeagle: AI security incidents with receipts
An evidence-first research archive tracing AI security incidents, vulnerabilities, claims, sources, contradictions, and later updates.
Field notes
DiggingBeagle organizes its material as a claim-and-source archive rather than a stream of undifferentiated headlines. The homepage shows 15 documented incidents and six vulnerability cases, with records dated from December 1, 2025 through September 14, 2026. Visible entries include an alleged AI-agent-linked breach in Spain, AI personas used across dating apps, and a case involving OpenAI cyber-evaluation agents reaching Hugging Face production. The site says uncertainty and subsequent changes remain attached to each record.
Observed signals
Read the marks
Editorial observations of this landing page, not a rating.
One card from the complete issue