Technology Apr 29, 2026 · 5 min read

We Built a 3-Layer Audit Trail (AI + GPS + Blockchain) to Eliminate Greenwashing in Ocean Conservation

Most environmental charities ask you to trust them. We built a system so you don't have to. I'm Valentin, co-founder of Thriving Planet Association, a Swiss nonprofit that recovers ocean plastic. But this isn't a fundraising post dressed up as a tech article. It's about a real engineering problem w...

DE
DEV Community
by Chad Buliamu
We Built a 3-Layer Audit Trail (AI + GPS + Blockchain) to Eliminate Greenwashing in Ocean Conservation

Most environmental charities ask you to trust them. We built a system so you don't have to.

I'm Valentin, co-founder of Thriving Planet Association, a Swiss nonprofit that recovers ocean plastic. But this isn't a fundraising post dressed up as a tech article. It's about a real engineering problem we had to solve: how do you make environmental impact verifiable, tamper-proof, and visible to anyone who funds it?
Here's what we built and why it matters for anyone working at the intersection of technology and social good.

** The Problem: "Trust Us" Doesn't Scale**

The environmental sector has a transparency crisis.
Donors are increasingly skeptical and rightfully so. Terms like "carbon offset" and "plastic neutral" have been so thoroughly abused by corporate greenwashing that many people have stopped believing environmental claims altogether.
For ocean conservation specifically, the problem is acute:

How do you prove that a specific kilogram of plastic was collected, rather than just estimated?
How do you ensure the GPS coordinates of a cleanup aren't fabricated?
How do you give a donor in Zurich verifiable proof of what happened on a beach in Southeast Asia?

We asked ourselves these questions before we ran a single collection. The answer was a 3-layer technical audit trail.

The Solution: AI + GPS + Blockchain

We call it the 3-Layer Audit Trail. Here's how each layer works:

** Layer 1 AI Verification (Weight & Classification)**

Every gram of plastic collected is weighed and photographed at the point of collection. A computer vision model classifies plastic by type (PET, HDPE, LDPE, etc.) and cross-references the weight against the visual volume of material. This eliminates human rounding errors and makes it significantly harder to inflate collection numbers.

Why this matters: Plastic classification data is also valuable for downstream recycling partners — knowing the composition of collected material improves recycling yield and creates a secondary accountability loop.

Layer 2 GPS Timestamping

Each collection event is tagged with:

Precise GPS coordinates (latitude/longitude)
Timestamp (UTC)
Collector ID
Photo evidence

This data is captured at the moment of collection and cannot be retroactively altered. Donors receive a digital certificate showing the exact location where their sponsored kilogram was recovered.
Why this matters: GPS coordinates are independently verifiable. Anyone can open Google Maps and see the collection site. There's nowhere to hide.

** Layer 3 Blockchain Immutability**

The AI classification result, weight, GPS coordinates, and timestamp are hashed and written to a blockchain ledger. Once written, this record cannot be modified, deleted, or backdated.
This gives donors something unprecedented in charitable giving: an immutable, publicly auditable record of their impact.

** What This Looks Like in Practice**

When a donor contributes CHF 100 to our campaign, here's what happens:

  1. Collection team arrives at verified site (GPS tagged)
  2. Plastic is collected and weighed on calibrated scales
  3. The AI model classifies and validates the weight/volume ratio
  4. GPS + weight + timestamp → hashed and written to blockchain
  5. Donor receives digital certificate with:
    • Collection site coordinates
    • Weight recovered
    • Blockchain transaction ID (publicly auditable)
    • Photo evidence from the collection No black box. No "trust us." Just verifiable data.

** Why We're Sharing This on Dev. to**

We're currently running a CHF 1,000 crowdfunding campaign to launch a new verified collection point. But more than the funding, we want to start a conversation in the tech community about what verified impact infrastructure should look like.
We believe the tools exist today — computer vision, GPS, blockchain — to make greenwashing technically impossible. Not harder. Impossible.
If you're a developer, data scientist, or blockchain engineer who's thought about applying your skills to environmental problems, we'd love to connect.
And if you want to support the campaign directly:

CHF 25 = 1kg of plastic recovered and verified, with full audit trail
CHF 100 = 4kg + a digital certificate with exact GPS coordinates of your impact

👉 Support the campaign on GoFundMe

**https://www.gofundme.com/f/thriving-planet-associations?lang=en_GB

Open Questions We're Still Working On

We're not claiming to have solved everything. Here are the hard problems we're still wrestling with:

The Oracle problem with blockchain is only as trustworthy as the data written to it. If a collector falsifies the weight before it's hashed, the chain records a false truth immutably. We're exploring IoT-connected scales as a mitigation.
Scalability writing every collection event to a blockchain has cost and throughput implications. We're evaluating layer-2 solutions and batch hashing approaches.
Accessibility blockchain certificates are meaningless to most donors without a readable interface. We're building a human-friendly dashboard that abstracts the technical complexity.

If you've worked on similar problems in supply chain verification, carbon accounting, or impact measurement, we'd genuinely love to hear from you in the comments.

** Final Thought**

The technology to make environmental impact fully transparent already exists. The question is whether organisations are willing to build the infrastructure to use it even when opacity is more convenient.
We chose transparency. And we're betting that the people who fund conservation work will choose it too.

Thanks for reading.

DE
Source

This article was originally published by DEV Community and written by Chad Buliamu.

Read original article on DEV Community
Back to Discover

Reading List