AIGIS

AIGIS protects kids online with real-time AI that detects threats like grooming, violence, and distress across chats, games, and media—ensuring a safer digital world for every child.

Comment

Inspiration

The rise of online threats targeting underage users—such as grooming, exposure to explicit content, deepfakes, and hate speech—demanded a smarter, proactive solution. We were inspired to build AIGIS to combine AI innovation with social responsibility, creating a system that not only detects dangers in real time but also empowers parents and educators to intervene effectively.

What it does

AIGIS is an AI-powered safety platform that protects minors online through multimodal threat detection. It analyzes conversations, media, and behavior across games, chat apps, and social platforms to detect grooming, fake identities, violence, nudity, offensive language, brainrot content, deepfake audio, and signs of emotional distress. It also generates dynamic reports for guardians and includes a conversational assistant (Aigisso) to interpret the findings.

How we built it

We used Python, PyTorch, ONNX, and Django for the backend, integrating models for vision, NLP, and audio processing. Video and browser extensions feed data into AIGIS for analysis. The frontend is built with modern web tools and integrates with Langchain and RAGFlow for generative reports and chatbot support. Development was done collaboratively via GitHub and VSCode.

Challenges we ran into

  • Handling real-time data across different modalities and formats
  • Integrating multiple AI models efficiently without bottlenecks
  • Balancing detection accuracy with user privacy
  • Limited access to child-safe, annotated datasets for sensitive topics
  • Ensuring cross-platform compatibility (web, extensions, video tools)

Accomplishments that we're proud of

  • Building a working end-to-end prototype capable of real-time threat detection
  • Successfully integrating multimodal AI (vision, text, audio, behavioral analysis)
  • Developing a generative reporting tool grounded in actual content
  • Creating Aigisso, a contextual AI chatbot that explains risks to guardians
  • Raising awareness about digital safety for minors through tech

What we learned

We gained experience working with AI safety models across domains, and learned how to build scalable, privacy-aware pipelines. We also learned how to bridge ethical considerations with technical constraints, and how to make AI interpretable for non-technical users.

What's next for AIGIS

  • Improve deepfake and behavioral pattern detection using larger datasets
  • Expand emotional distress analysis with physiological signal tracking
  • Deploy a public beta with parental dashboards
  • Partner with schools and NGOs for testing and impact measurement
  • Add multilingual support and platform-specific integrations (Discord, Roblox, etc.)

What to Submit

Chosen Track: 2
Project summary: AIGIS is an AI-powered safety platform that safeguards underage users online through real-time multimodal threat detection. It analyzes conversations, media, and behavior across chat apps, games, and social platforms to detect grooming, fake identities, violent or explicit content, and emotional distress.
Designed for families, integrated into browsers, and built with modern web technologies. AIGIS bridges innovation and social impact.

Dataset links:

Built With

Share this project:

Updates