The archive · Product Ideas · Product decision · 2025
WWF's GhostNetZero trains an AI on sonar so anyone can help haul ghost nets from the sea
A custom AI learns to read sonar for lost fishing nets, then every sighting is published on a playful platform for the world to act on.
WWF · Accenture Song Germany
What it had to solve
Lost or abandoned fishing gear — 'ghost nets' — accounts for up to 30% of the plastic in the sea and keeps damaging marine life. WWF had spent years on extensive manual search, which could only ever act locally.
How it works
GhostNetZero tackles a slow, hidden environmental problem: lost or abandoned fishing gear, called ghost nets, can account for up to 30% of the plastic in the sea and keeps damaging marine life long after it is lost. For years the WWF hunted them through extensive manual search, which worked locally but could never match the scale of the problem.
The creative leap was to split the task. A custom-built AI learned to detect ghost nets from sonar data, doing the seeing that human search teams could not scale. The findings are then published via a platform and app designed to be easy to use and enjoyable, so the data becomes a playful experience rather than a spreadsheet.
The result reframed conservation work as a global, participatory loop: collaborators around the world can take action on a detected net wherever it is found. The project is archived in D&AD's Future Impact section as an example of scaling impact from local action to global collaboration.
Why it lands
- Ghost nets are invisible underwater, so the bottleneck was detection, not willingness to clean up.
- Manual search had already proven WWF's commitment for years, and shown that human effort alone could not scale.
- Publishing machine-found targets turns passive supporters into people who can act on a specific, real net.
- A playful interface lowers the barrier for a global audience that would never join a formal search team.
What it did
WWF's hunt scaled from a local, manual effort into a global collaboration: once the AI finds a net, anyone on the platform can take action instead of waiting for a search team.
What you can take
When scale is the bottleneck, split the job: let a machine read the signal at global volume and let the public own the response, with an interface that makes action feel like play.
Since then
GhostNetZero reframed WWF's ghost-net work from a search operation into an open platform: the AI supplies the detections and the app supplies the action, so responding to a net no longer depends on where a WWF team happens to be. Its archive placement in D&AD's Future Impact category positions the model as a template for pairing machine detection with public participation.
Sources
spotted an error? The archive wants to know.
Your turn
You just read one. Describe the brief you are staring at, and see who has been given the same problem.
Free account · 3 free questions · no card