Real-time Analytics
WWF Germany needed a faster, more reliable way to find ghost nets hidden in large volumes of archived sonar data because manual review was slow, costly, and too inconsistent to support large-scale recovery operations.
Reported impact
WWF Germany reduced the labor needed to inspect sonar data, improved detection accuracy, and created a scalable collaboration model that helps turn previously unused survey data into actionable recovery work for ghost-net removal.
WWF Germany partnered with Accenture and Microsoft's AI for Good Lab to create GhostNetZero, an AI-powered platform that turns raw sonar scans into usable recovery targets for ghost-net removal teams. Sonar data from scientific surveys, seabed mapping, and navigation programs is uploaded into the platform instead of being left in archives. AI models scan the files, identify suspicious patterns that may indicate lost fishing gear, and flag candidate targets for marine experts to validate. Confirmed locations are then logged, mapped, and shared through a dashboard that expands as new data donors contribute files across countries. The system creates a shared knowledge base of underwater hazards while keeping participation simple through secure architecture and live dashboards. By combining sonar ingestion, AI detection, expert validation, and shared mapping, GhostNetZero shifts ghost-net recovery from manual footage review to a repeatable digital workflow that is easier to scale region by region.
Research and survey organizations upload sonar scans to GhostNetZero, where AI models analyze files and flag likely ghost-net targets. Marine experts validate those hits, and confirmed targets are logged and mapped in the platform dashboard so recovery teams and new partners can act on the growing shared data set.
"AI scans files, flags suspicious areas, and sends them to marine experts for validation."
"Confirmed targets are logged and mapped, building a shared knowledge base of underwater hazards."
"GhostNetZero—an AI-powered platform that transforms sonar scans into precise recovery targets"
"~95% detection accuracy in AI model precision in identifying ghost nets"
Review how this reference was scored and trace each normalized claim back to its documented source.
Scoring methodology
Customer Reference Value measures usefulness as a customer reference. Evidence Confidence measures the quality and traceability of available evidence, not independent verification.
Traceability
Field coverage shows which facts each source supports. Dates are shown only when documented by the source.
The report states: 490 Menschen beschäftigte der WWF Deutschland im Durchschnitt des Geschäftsjahres 2022/2023; the standard mapping classifies 250-999 employees as Mid-market.
Related
Anthropic Sample Archived · Oct 4, 2026 International Rescue Committee IRC teams needed to analyze frontline health data quickly enough to influence staffing, outreach, and supplies, but the information came from ministry-run facilities in inconsistent paper and Excel formats and often had… 4+ hours per month saved by half of surveyed pilot users Claude Enterprise Anthropic Sample Archived · Sep 28, 2026 International Rescue Committee IRC wanted proof that AI could turn inconsistent frontline health data into useful local decision support despite electricity constraints, poor connectivity, and ministry-defined indicators that varied across facilities. 2,700+ health facilities across 40 countries Claude Enterprise 
Google Cloud Sample Archived · Sep 17, 2026 IATA IATA wanted a repeatable short-term air-travel forecasting signal to complement long-term econometric models, but needed the entire workflow to run automatically each month with richer features and no manual… One global ML model produces approximately 5,700 bidirectional forecasts Cloud KMS · Cloud Run functions · Cloud Scheduler 
