Asilla's camera AI may have prompted two suicide interventions

AI system

System: Asilla AI behavior-analysis system for security cameras
Model: Not assessed
Release: Not assessed
Application developer: Asilla Inc.
Role: Not assessed
Configuration: Not assessed
Inputs and outputs: Not assessed
Autonomy: Not assessed
Workflow: The system analyzed security-camera footage for behavior that suggested a risk of jumping. Guards or other staff responded to its alerts. [1] [2]
Human oversight: Not assessed
Relation to AI: Kyodo attributes both interventions to Asilla’s account. The company’s original product documentation supports the existence and intended mechanism of jump-risk alerts, but does not independently verify either intervention. The facilities, dates, alert logs, and subsequent outcomes are not supplied. [1] [2]
AI contribution: Asilla says that its camera system generated alerts before two interventions. Its product documentation describes how alerts work, but the company supplies no dates, locations, or alert logs for these incidents. [1] [2]

Outcome identifiers: Not assessed
Benefit: Asilla reported two interventions after camera alerts. It did not identify the facilities or dates. [1] [2]
Category: Improved Physical Health
Status: Not assessed
Beneficiary groups: People. Organizations.
Beneficiaries: Not assessed
Reach: Not assessed
Distribution and access: Not assessed
Duration: Not assessed
Expected?: Yes
Costs and harms: Not assessed

Studies: Not assessed
Study design: Not assessed
Measurement: Not assessed
Setting and sample: Not assessed
Measures: two reported interventions after alerts; dates and facilities undisclosed
Comparison: Not assessed
Results: two reported interventions after alerts; dates and facilities undisclosed
Additional value from AI: Not assessed
Limitations: Not assessed
Denominators: Not assessed
Materials for replication: Not assessed

Participants: Developer: Asilla Inc. Operator: About 40 railway stations and commercial buildings in Japan (operators not named).
Jurisdictions: Developer: Japan.
Ownership and control: Not assessed
Dependencies: Not assessed
Cross-border connections: Not assessed
Geopolitical implications: Not assessed

Conditions for replication: Staff presence to act on alerts. Acceptable false-alert rates for operators. Camera coverage of the risk zones.
Replication attempts: Not assessed
Later findings: The retained sources do not establish subsequent outcomes for the two people. [1] [2]
Availability: Not assessed
Institutional decision: Not assessed
Grounds: Not assessed
Action and result: Not assessed
Success criterion: Not assessed

Source numbers: AIBD-0234-source-0001–0002 [1] [2]
Citations: Numbered references below. [1] [2]
Language: Not assessed
Source relationships: Independent, Developer [1] [2]
Original evidence: Kyodo reports Asilla’s account of two interventions. The complete Straits Times republication is available, but the original distribution page is inaccessible. A separate Osaka report describes false alarms and gaps in response, without identifying either of these incidents. [1] [2]
Access and versions: The complete Kyodo report is available through The Straits Times. Chiba TV identifies its original distribution link, which is inaccessible. Asilla’s original product release is available. [1] [2]
Supported findings: Not assessed
Evidence links: Not assessed
Evidence strength: Not assessed
Missing evidence: Not assessed

Assessed benefits: People. Organizations.
Depth: People: 3. Organizations: Not assessed.
Reasons: People: Two people in apparent crisis were reached by guards after alerts, per the developer; no independent account of either case. Organizations: No operator-reported outcome or false-alert data in the sources.
Observed or intended: Not assessed
Rubric: Not assessed
Assessor and date: Not assessed
Alternative assessments: Not assessed
Fill: Independently reported

Kind: Not assessed
Related entries: Not assessed
Field: Public safety / suicide prevention. Transport and real estate.
Purpose: Detect behavior indicating suicide risk (and illness, immobility or violence) in security-camera footage and alert staff.
Dates: Source published: 1 June 2026. Updated: 24 October 2026. [1] [2]
Stage: Not assessed
Version: SHA-256: f901188e7fee…

News report Independent author

Kyodo News via The Straits Times, 1 Jun 2026

Kyodo’s report is original journalism carried by The Straits Times. The two intervention accounts are explicitly attributed to Asilla, not independently confirmed by the reporter.
Press release Developer

Asilla, “アジラ、最先端のAI警備システム『AI Security asilla』をバージョンアップ。ver.2.4のリリースを開始！,” January 26, 2024, “飛び降り予兆検知機能.”

Asilla’s January 2024 release identifies the product function and technical inputs. It is separate primary documentation of the mechanism, not corroboration of the two events.