VertoAI

Inspiration

Misinformation spreads fastest on WhatsApp, but most fact-checking tools exist outside it. In India especially, people forward messages based on trust, urgency, or fear — not because they intend harm, but because verifying information is inconvenient.

We noticed that even when reliable fact-checks exist, they are rarely used because users don’t want to leave WhatsApp, open a website, or understand technical explanations.

VERTO AI was inspired by a simple question: What if fact-checking worked exactly where misinformation spreads, and required no technical knowledge at all?

What it does

VERTO AI is a WhatsApp-first misinformation detection system.

Users simply forward any message, image, or video to VERTO AI on WhatsApp, and within seconds receive a clear, human-readable verdict:

❌ False

⚠️ Suspicious

✅ Verified

The response is delivered in simple Hindi or English, with clear guidance such as “Do not forward” or “This information is safe.”

Behind the scenes, VERTO AI:

Detects semantic variations of the same false message

Identifies manipulated or AI-generated images and videos

Tracks how fast misinformation is spreading

Prioritizes viral threats before they escalate

An optional mobile app provides added features like:

Past verification history

Trending misinformation alerts

Family safety summaries

Emergency “panic check” during crises

The core verification, however, works entirely inside WhatsApp.

How we built it

VERTO AI uses a multi-modal AI architecture designed for real-time analysis:

Text Analysis: Semantic embeddings are generated to detect different versions of the same claim, even if wording changes.

Image & Video Analysis: Media is scanned for manipulation, deepfake indicators, and audio-visual mismatches.

Virality Tracking: Similar reports are grouped and counted to detect rapidly spreading misinformation.

Human-Readable Translation Layer: Technical AI outputs are converted into simple, calm explanations suitable for non-technical users.

The backend is built with:

FastAPI for APIs

MongoDB (including vector search) for claim tracking

Cloud-based AI models for text, image, and video analysis

The frontend app is intentionally minimal and designed for 35+ users, focusing on clarity and trust.

Challenges we ran into

Designing for non-technical users: Explaining AI decisions without showing technical details required careful UX and language design.

Balancing speed and accuracy: Real-time responses were essential, but false positives could damage trust.

Handling multiple content types: Text, images, and videos all require very different analysis pipelines.

Cost and scalability considerations: AI-based media analysis is resource-intensive and required smart prioritization.

Trust and responsibility: Labeling content as “false” has social and ethical implications, so confidence and wording had to be handled carefully.

Accomplishments that we're proud of

Built a WhatsApp-first system instead of a traditional app-first approach

Successfully unified text, image, and video analysis in one pipeline

Designed a UX that works for 35+ non-technical users

Implemented virality tracking, not just one-off fact checks

Created a system that prevents misinformation before it spreads, not after

What we learned

Users don’t want technical explanations — they want clear guidance

Trust is built through simplicity, tone, and consistency, not accuracy numbers

Misinformation is a behavioral problem, not just a technical one

Prevention is far more effective than correction

AI systems must be designed with human psychology in mind

What's next for VertoAI

Our next steps include:

Expanding early-warning alerts for rapidly spreading misinformation

Improving regional language support, starting with Hindi and English

Adding human-in-the-loop review for sensitive or uncertain cases

Partnering with NGOs, journalists, and civic organizations

Strengthening privacy, transparency, and auditability

Our long-term goal is to make VERTO AI a quiet, trusted safety layer that protects people from misinformation — without changing how they use WhatsApp.

Built With

• flutter
• react