Drink Spiking Detection
Archangel AI detects drink spiking attempts in real time by analysing hand movements near unattended drinks, hovering behaviour, and suspicious timing patterns across your venue.
How It Works
Under the hood
Archangel's drink spiking model watches for a specific set of behavioural patterns that precede spiking attempts. The system tracks hand proximity to unattended glasses, identifying when someone reaches toward a drink that isn't theirs. It flags hovering behaviour where a person lingers near drinks without picking one up or engaging with anyone at the table.
The model also analyses timing patterns. A hand that passes over a glass too slowly, or a person who returns to the same spot near unattended drinks multiple times, triggers escalating confidence scores. These scores combine with contextual signals like time of night, crowd density, and whether the drink owner is absent.
When the system reaches a confidence threshold, it sends an instant alert to bar staff or security with the camera feed, timestamp, and location. Staff can respond before a spiked drink is consumed. Every alert is logged with full metadata for incident review and duty-of-care records.
Why It Matters
The real-world impact
Drink spiking is one of the most underreported crimes in nightlife. Victims often don't realise what happened until it's too late, and traditional CCTV only helps after the fact. There's no prevention, just footage to review after someone has already been harmed.
Archangel shifts the model from reactive to preventive. By detecting the behaviour before the drink is consumed, venues can intervene in the moment. This protects patrons directly and gives venues a documented, proactive safety record that stands up to regulatory scrutiny and insurance claims.
<2s
Detection Speed
94%
Incident Prevention Rate
<3%
False Positive Rate
Capabilities
What it detects
Purpose-built detection models tuned for real-world conditions and operational environments.
Hand Proximity Tracking
Monitors hand movements near glasses and containers, distinguishing between normal drink handling and suspicious reach patterns toward unattended beverages.
Hovering Behaviour Analysis
Identifies individuals who linger near drink areas without social engagement, flagging patterns consistent with pre-spiking surveillance behaviour.
Temporal Pattern Recognition
Tracks repeated approaches to the same area, unusual timing of hand movements over drinks, and correlates these with the absence of the drink owner.
Instant Staff Alerts
Pushes real-time notifications to bar staff and security teams with camera feed, location pin, and confidence score so they can intervene before consumption.
Sectors
Where it applies
This detection capability is built for the environments where it matters most.
See it in action
Book a demo to see how drink spiking detection works with your existing cameras. No new hardware required.
Free consultation. Works with any CCTV system. Live in under 48 hours.