Project Description
Clarity – Building Digital Confidence Online
Clarity is a browser extension that helps users safely navigate the web by detecting scams, fact-checking claims against government data, and explaining risks in plain English. Designed especially for vulnerable groups, it builds digital confidence through trust, protection, and accessible learning.
Submission Video
(Coming soon!)
The Problem
The internet powers communication, services, and community — but it’s also filled with risks:
- Phishing scams trick people into giving away their accounts.
- Fake ads lure users into downloading malware.
- Misinformation spreads faster than facts, eroding public trust.
- Vulnerable groups — children, seniors, and digitally inexperienced users — are most at risk.
Without the right tools, these users are left exposed and unable to participate confidently in digital spaces.
The Solution: Clarity
I built Clarity to give users digital confidence: a browser extension that detects cyberthreats, cross-checks claims against government datasets, and explains the findings in plain, human language through an integrated assistant.
Misinformation Claim Detector
Clarity compares claims in online articles against trusted government datasets (e.g., ABS housing data).
When misinformation is detected, users see a clear flag and can explore the contradiction. This helps fight disinformation while empowering users with trusted, official evidence.
Cyberthreat Detector
Clarity scans pages for phishing phrases, suspicious domains, and scammy ads or popups.
Users get real-time alerts before they click something harmful, reducing risks of fraud, ransomware, and identity theft.
Chatbot Assistant ("Clarity")
Many warnings are hard to understand. The assistant takes the analysis (detected threats, misinformation checks, contradictions) and explains them in simple English.
Users can ask:
- “Why is this article misleading?”
- “Is this link safe?”
- “What does the official data say?”
The assistant turns complex findings into casual, everyday language — making online safety and learning accessible to everyone.
Technical Details
- Browser Extension – Injected content script for real-time scanning.
- Local Model Integration – Uses a GGUF model with CUDA acceleration via
llama.cpp for natural language explanations.
Datasets Used – Australian Bureau of Statistics (ABS) housing data.
This dataset is used as a proof of concept for the misinformation detection feature.
For example, when an article claims that housing prices are falling, Clarity checks this against official ABS data to highlight contradictions.
While the current implementation uses a single dataset for demonstration, the design is scalable: additional government datasets (e.g., health, education, employment) can be integrated to broaden coverage and further strengthen digital trust.
Privacy-First – All analysis is done locally; no external API calls.
Challenge Fit
Clarity directly addresses the GovHack 2025 challenge: Digital Confidence – Tools for Safe Online Participation by combining practical threat detection with government-backed fact-checking.
- Detection & Protection – Real-time scanning flags phishing phrases, suspicious links, and scammy popups to prevent harm before it happens.
- Trust & Integrity – Misinformation analysis is grounded in official ABS data, ensuring that online claims are verified against trusted government sources.
- Inclusion & Protection of Vulnerable Groups – Clarity is built for users who are most at risk — children, seniors, and those less digitally experienced — by explaining findings in plain, everyday language.
- Digital Literacy & Empowerment – The integrated assistant transforms technical warnings into simple, actionable advice, empowering users to understand risks instead of just fearing them.
- Scalability & Future Growth – While demonstrated on housing data, Clarity is designed to integrate multiple government datasets, making it adaptable to diverse scenarios and communities worldwide.
By grounding online safety in official data and accessible explanations, Clarity helps restore digital confidence and ensures that all users can participate securely and meaningfully in online spaces.
Moving Forward
- Expand misinformation checks with additional government datasets.
- Broaden cyberthreat detection to include real-time feed updates.
- Enhance the chatbot with multilingual support for inclusivity.
- Develop Clarity into a fully production-ready browser extension that can be installed and used by anyone.
Project Repository
GitHub: Clarity – Digital Confidence Assistant
Data Story
To demonstrate misinformation detection, I used multiple datasets from the Australian Bureau of Statistics (ABS) related to housing and household spending. These datasets were chosen because housing costs and affordability are common topics for misinformation and misleading claims online.
Datasets Used
- Estimates of Household Spending, Australia – Used as part of the misinformation feature to validate claims about consumer spending trends.
- Income and Work Census – Used to cross-check claims relating to income levels, employment, and affordability.
- Housing Cost – Used to verify statements around housing price movements, rental costs, and household financial stress.
How the Data Was Used
- The datasets (provided in Excel
.xlsx format) were converted into JSON for integration into the browser extension.
- A custom script was built to correlate, aggregate, and summarise key indicators from the datasets.
- The script then generated trimmed and fact-checked summaries, which are used as the reference base for detecting mismatches in online articles or webpages.
- Example: If an article claims “Housing prices are crashing”, Clarity checks the claim against the latest ABS housing cost data to highlight any contradiction.
- These summaries are also passed into the assistant context, enabling the chatbot to explain official ABS findings in plain English alongside the flagged misinformation.
Why It Matters
This workflow ensures that misinformation detection is grounded in trusted, official government data, rather than arbitrary sources. While the proof of concept currently uses housing-related datasets, the process is reusable and scalable: the same scripts can be applied to other ABS datasets (health, education, employment, etc.) to expand Clarity’s fact-checking coverage.
