AETHRA AI

Aethra — The Bridge Between AI Transparency and Human Well-being.

Comment

Inspiration

Inspiration

We were inspired by the increasing number of people impacted by AI-generated misinformation, deepfakes, and unethical data collection. While AI is transforming our world, it often lacks human-centered ethics and emotional consideration for its victims.

We envisioned Aethra as more than a detection tool — it’s a compassionate companion that brings together:

AI ethics awareness 🧭

Transparency in digital behavior 🔍

Mental health support for AI victims 💬

“Aethra” (meaning bright or pure air in Greek) symbolizes clarity — our goal is to bring clarity to the ethical fog surrounding AI.

What it does

What It Does

Aethra is an AI-powered web platform that combines ethical intelligence and emotional support in one space.

It performs three main functions:

🧠 AI Ethics Detection

Users can upload text, images, or videos.

Aethra’s AI analyzes the content and determines whether it contains unethical elements, such as:

misinformation,

deepfakes,

privacy violations, or

harmful AI-generated content.

It then explains why the content may be ethical or unethical, promoting awareness and digital literacy.

💬 Aethra Care (Counseling Chat)

An AI-driven chat assistant offers emotional counseling and guidance for people affected by unethical AI use or data loss.

It provides empathetic support, recovery tips, and safety resources — blending emotional intelligence with AI understanding.

🌍 Global Ethics Visualization (3D Globe)

Users can explore a 3D interactive Earth showing country-level AI ethics insights.

Clicking a country displays its states and example unethical sites or incidents.

It helps users visualize how AI misuse or misinformation spreads worldwide.

🧾 Data Misuse Reporting

Victims or witnesses of unethical AI behavior can submit detailed reports about their data loss or misuse.

These reports help build a database for ethical monitoring and future prevention.

How we built it

How We Built It

Aethra’s architecture combines advanced AI tools and modern web technologies:

🏗️ Frontend

Next.js (React Framework) for building a smooth, interactive interface.

React-Globe.GL + Three.js to create a 3D interactive Earth, allowing users to select countries and view AI ethics data.

TailwindCSS for futuristic dark UI styling and responsive design.

Integrated support for file uploads (text, image, video) for analysis.

🧠 Backend

Node.js + Express server handling API routes:

/analyze — Processes user text/image/video and predicts ethical rating.

/chat — Routes counseling messages to an LLM (e.g., Gemini or GPT API).

/report — Logs data misuse reports securely in a database (MongoDB planned).

AI Model pipeline (conceptual):

𝑓 ( 𝑥

)

softmax ( 𝑊 ⋅ BERT ( 𝑥 ) + 𝑏 ) f(x)=softmax(W⋅BERT(x)+b)

where 𝑓 ( 𝑥 ) f(x) classifies input 𝑥 x as ethical, unethical, or needs review.

🪐 Visualization

Countries and states are dynamically rendered using GeoJSON data.

Clicking a country displays its states and “flagged websites” (demo data or from API).

Fully optimized with lazy loading and data fallback for performance.

Challenges we ran into

Challenges We Faced

3D Globe Performance Issues: The initial version of the globe was too heavy for browsers — we optimized it using React-Globe.GL and reduced polygon data for smoother rendering.

LLM Integration Limits: Managing long AI conversations and maintaining contextual empathy required careful API handling and fine-tuned prompts.

Ethics Classification Dataset: Finding and preprocessing real “ethical vs. unethical” AI content datasets was challenging, as most data is domain-specific or unstructured.

UI Consistency Across Devices: Ensuring the futuristic design stayed responsive and smooth on all devices took multiple iterations.

Accomplishments that we're proud of

Developed AI-powered content analysis

Implemented an AI text, image, and video analyzer that identifies unethical patterns like deepfakes, misinformation, and privacy violations.

Integrated clear explanations to help users understand why content is flagged.

💬 Introduced “Aethra Care” — an empathetic counseling chat

Built a conversational AI counseling assistant for victims of data misuse and emotional stress.

Focused on ethical empathy — giving AI a human touch.

🧾 Created a secure reporting system

Designed a form for users to report data loss or unethical AI incidents, enabling community-driven awareness.

Ensured structured input to support backend data collection and future policy analysis.

🎨 Designed a unified, futuristic user interface

Crafted a sleek, glowing dark-mode UI symbolizing the balance between technology and transparency.

Every page — detection, chat, report, and globe — fits into one consistent ecosystem.

🌍 Brought ethics, awareness, and empathy together

The proudest part: blending AI ethics detection and mental well-being support into one purposeful solution — something rarely seen in AI-focused projects.

What we learned

What We Learned

Throughout development, we learned how to:

Integrate AI ethics concepts into practical, user-facing tools.

Combine LLM-based conversational AI with traditional ML analysis pipelines.

Use 3D data visualization to communicate complex ethical issues effectively.

Balance technical accuracy with human empathy in AI-driven design.

We also explored how ethical frameworks like the EU AI Act and UNESCO’s AI Ethics Guidelines can be reflected in real applications

What's next for AETHRA AI

The Vision Ahead

Our future roadmap includes:

Real-time global AI ethics dashboard integrated with verified reports.

Expansion of counseling chat with emotional sentiment detection.

Partnership APIs for NGOs and digital safety organizations.

User community for AI ethics education and awareness.

Our mission: To make AI transparent, ethical, and emotionally aware. Our mission: To make AI transparent, ethical, and emotionally aware.

Built With

  • 11:48?pm]-millan-sabu:-ai-&-ml-layer-gemini-pro-?-ethical-scoring-gemini-vision-?-deepfake-detection-vertex-ai-sdk-?-fairness-&-explainability-optional-cnn-fallback-models-datasets:-faceforensics++
  • 15
  • aif360-[11/8
  • analytics
  • api
  • authentication
  • axios
  • backend
  • bigquery
  • cloud
  • css
  • database
  • dfdc
  • express.js
  • firebase
  • firestore
  • framer
  • frontend
  • google
  • mongodb
  • mongoose
  • next.js
  • node.js
  • react)
  • shadcn/ui
  • stack
  • tailwind
Share this project:

Updates