AI Helps South Korea Stop 99% of Suicide Attempts on Han River Bridges in Seoul

Why It Matters

The high survival rate shows AI can dramatically improve public‑safety outcomes, offering a replicable model for cities confronting similar mental‑health crises.

Key Takeaways

• AI monitors 900 cameras on 17 Han River pedestrian bridges.
• System triggers alarm after 300 seconds in loiter zones.
• 1,270 suicide attempts recorded last year, 99% survival rate.
• Only 10 deaths occurred despite over 1,000 attempts.
• Model may inspire global suicide-prevention tech deployments.

Pulse Analysis

South Korea has long grappled with one of the world’s highest suicide rates, and the capital’s Han River bridges have become a stark symbol of the crisis. City officials report more than 1,000 bridge‑related attempts each year since 2022, with 1,270 dispatch calls recorded in 2025 alone. While the raw numbers are sobering, the survival rate has climbed dramatically, reaching 99 percent after the introduction of a dedicated AI‑driven monitoring centre in 2021. The initiative reflects a broader shift toward data‑centric public‑health interventions.

The Hangang Bridge CCTV Integrated Control Centre operates a network of 900 cameras spanning 17 of the 21 pedestrian‑accessible bridges. Advanced computer‑vision algorithms scan each feed in real time, flagging any individual who lingers more than five minutes in predefined ‘loitering zones’. When the AI detects such behavior, an audible alarm prompts officers to intervene on the spot, often before the person reaches the edge. Since deployment, the centre has responded to three‑four suspected attempts daily, converting 1,260 of those encounters into successful rescues.

The program’s success has sparked interest from municipalities worldwide seeking scalable, technology‑enabled solutions to mental‑health emergencies. By coupling AI precision with human judgment, Seoul demonstrates how low‑latency alerts can augment traditional crisis‑intervention teams without replacing them. Nonetheless, privacy advocates caution that continuous facial‑recognition surveillance must be governed by transparent policies and strict data‑retention limits. As other cities evaluate similar deployments, the balance between public safety gains and civil‑liberties safeguards will shape the next generation of AI‑assisted emergency response. Early pilots in Tokyo and Toronto are already testing comparable frameworks.