Oxford AI for Good Hackathon challenges students to build AI solutions for social good

Held at Reuben College on 12–13 June 2026, the 24-hour hackathon brought together 75 participants to develop technology addressing some of society’s most pressing challenges. Sponsored by EnSpire Oxford, Claude Builder Club @ Oxford and Canva, the event offered four tracks: healthcare, life sciences, climate change, and values & society.

The AI4DH Lab team – Zi Cai, Noura Aqeeli, Chenqi Li, Joy Lai and Anna-Maria Geist – selected the values & society track. They focused on the often unseen human cost of online content moderation, where moderators are exposed to a constant stream of violent and graphic material on behalf of online platforms.

"Content moderation is at a breaking point", the team said. As malicious users increasingly exploit AI to generate harmful content and evade detection, human moderators face growing workloads alongside increasing legal and regulatory pressure.

The team developed a hierarchical AI pipeline designed to filter and abstract harmful content before it reaches a human reviewer. The approach keeps people involved in the decision-making process while aiming to reduce the psychological toll of repeatedly viewing disturbing material.

Using open-source vision-language models alongside Claude Code, Claude Design and Vercel, the team built a live prototype during the hackathon. They also developed a business case for the solution, including a business model and go-to-market strategy, before presenting their concept in a pitch lasting less than five minutes.

Projects were judged by domain experts on their potential impact, innovation, technical quality, feasibility and presentation. The AI4DH Lab team was awarded second place and a £200 prize.

The team’s project is featured in the AI4DH Lab newsletter.