Publisher: Project WOODPECKER (haveibeenflocked.com)
Relevant To: ALPR presentation (August) — IMPORTANT: this is an inference, not a confirmed fact
Verification Status: Verified — Content Confirmed, Explicitly Labeled as Inference by Source Itself
Notes: Confirmed via screenshot. Critical nuance: this specific content sits under the page’s “INFERENCES – Latent Hardware Capability” section, NOT its “FACTS” sections — the source itself explicitly distinguishes these. The evidence cited is a code excerpt from Flock’s own camera-management JS bundle (source: camera-management/static/js/main.8d7d9361.js:4991), showing an enum of detection/alarm types that includes MOTION_DETECTION, AUDIO_DETECTION, POS_INFO, SCHEDULED_RECORDING, CROSS_LINE_DETECTION, and FACE_DETECTION (value 8) among others. This shows a “FACE_DETECTION” constant exists in Flock’s client-side code — likely inherited from an underlying/white-labeled camera platform — which the site’s authors treat as evidence of latent hardware capability, NOT proof that Flock’s product actively performs or exposes facial recognition to customers. This directly complicates (but does not straightforwardly refute or confirm) Flock’s public claim that “Flock cameras do not use facial recognition technology” (flocksafety.com/trust). For BPA materials: cite this precisely as “code evidence of a FACE_DETECTION constant in Flock’s software, which the source characterizes as an inference of latent capability, not confirmed active use” — do not overstate this as proof Flock performs facial recognition, per BPA’s no-hedged-citation standard.