The Problem
Wildlife exploitation
has a platform problem.
Over 550,000 posts on a single platform depict wildlife in tourist entertainment contexts. Most come from venues associated with animal cruelty. Social media has become a primary marketplace for wildlife trafficking — and platforms are not doing enough.
NSFA AI is a multimodal content moderation system built to change that. It combines computer vision, natural language processing, and LLM reasoning to detect animal exploitation at scale.
Platform Accountability
They have the tools.
They choose not to use them.
The world's largest social media platforms possess the technical capability to detect wildlife exploitation at scale — yet content that fuels trafficking and animal suffering continues to be hosted, amplified, and monetised.
NSFA AI exists to make a simple point: detecting this content is not hard. This entire system was built by a small independent initiative in a matter of days, using commercially available AI. If we can do it, platforms with billions in resources certainly can too.
The absence of wildlife exploitation as a recognised moderation category is not a technical limitation — it's a choice. This project is here to raise awareness of exactly that.
What you get
Everything you need to detect wildlife exploitation
Live Analysis Tool
Upload an image or paste a social media URL (TikTok, Instagram, YouTube). The system analyzes in real-time and returns a full risk profile.
Open the DemoModeration Interface
Human-in-the-loop review queue. Open any flagged item to inspect the full risk breakdown, then approve, restrict, or escalate.
Open ModerationIntegration Approach
How NSFA AI works and how platforms, NGOs, and researchers can use its structured output. Pipeline, schema, and integration paths.
View IntegrationScientific White Paper
Full methodology: risk scoring rubric, multimodal architecture, facility intelligence, evasion tactics, policy recommendations.
Read White PaperAbout the Initiative
Why NSFA AI exists, who created it, and the partners behind it — Wild Around and Magnetic Core Nature Lab.
About NSFA AIAccess by Request
NSFA AI is a hosted proof-of-concept. Partner platforms, NGOs, and researchers can request access to pilot the live system.
Get in Touch
Evasion Detection
Traffickers hide
in plain sight.
47% of wildlife selling posts use code words instead of species keywords. Sellers replace text with animal emojis 🦜🐍🦎, redirect to WhatsApp, and use "DM for details" to evade filters.
NSFA AI detects all of it — evasion tactics, transaction intent, and caption sentiment signals per analysis.
System Architecture
How NSFA AI Works
Input
Image upload or social media URL (TikTok, Instagram, YouTube, Facebook, X)
Vision Module
Species ID, environment classification, human detection, interaction type analysis
Context Module
Caption NLP, poster credibility, facility lookup, evasion & trafficking language
LLM Reasoning
Multimodal chain-of-thought synthesis → composite risk score (0–1) + 30+ fields
Wildlife Products Detection
Ivory. Rhino horn.
Pangolin scales.
NSFA AI identifies specific wildlife-derived products and returns a structured list — enabling granular conservation policy responses.