Artificial intelligence is now fighting gender-based violence in South Africa through a new policy playbook released by Stellenbosch University. The FemAI Policy Playbook came after President Cyril Ramaphosa declared gender-based violence and femicide (GBVF) a national disaster on November 20, 2025. This playbook shows how artificial intelligence can help government departments track cases, identify patterns, and protect survivors across the country.
The FemAI initiative is a collaboration between Stellenbosch University’s Policy Innovation Lab, Women Political Leaders (WPL), and Deutsche Gesellschaft für Internationale Zusammenarbeit (GIZ). The project is part of a broader series across selected African countries designed to advance AI-informed policymaking.
South Africa records some of the highest rates of violence against women globally. Between July and September 2024 alone, 957 women were murdered, 1,567 survived attempted murders, and 14,366 experienced serious assaults. The country’s First National Gender-based Violence Study revealed that more than one in three women have experienced physical or sexual violence.
The new playbook provides government officials with practical steps to use artificial intelligence effectively while protecting survivors’ privacy.
AI Tracks Gender-Based Violence Cases Across Government Systems
The FemAI Policy Playbook outlines how artificial intelligence can connect fragmented data across police stations, hospitals, and courts. Currently, these departments work separately. A survivor might report to police, seek medical care, and attend court proceedings, but no single system tracks their entire experience.
Dr. Itai Makone, lead author from Stellenbosch University’s Policy Innovation Lab, explains that artificial intelligence can integrate these separate databases. The technology would allow officials to follow a case from initial report through final resolution. This prevents survivors from repeating their stories at every stage.
The playbook describes four types of artificial intelligence for different tasks. Supervised learning can automatically sort police reports and witness statements into categories. Unsupervised learning analyses community safety reports to spot emerging trends. Reinforcement learning powers digital tools that guide survivors to appropriate legal, medical, or counselling services. Generative artificial intelligence helps policymakers identify patterns across thousands of reports.
Machine Learning Identifies Hidden Patterns in Gender-Based Violence Data
South Africa’s gender-based violence problem extends beyond what statistics show. Many incidents go unreported. DNA testing backlogs stretch for months. Police stations often lack basic equipment like rape kits. Some rural areas have limited access to reporting mechanisms.
Artificial intelligence can help address these gaps. The technology can analyse existing reports to identify geographic hotspots, time patterns, and risk factors. This information helps officials allocate resources more effectively.
The Policy Innovation Lab developed the playbook after convening political leaders, women entrepreneurs, academics, and civil society groups. These stakeholders identified practical barriers to implementing artificial intelligence solutions. They emphasised that technology alone cannot solve the crisis without proper funding and coordination.
Prof. Willem Fourie, who leads the Policy Innovation Lab, notes that participants at the G20 Social Summit stressed one recurring problem. Government departments must break down information silos to share data effectively. Without budget allocation for these technological tools, policies remain symbolic.
Artificial Intelligence Must Protect Survivor Privacy While Fighting GBV
The playbook addresses serious ethical concerns about using artificial intelligence in gender-based violence cases. Survivor data is extremely sensitive. Improper handling could expose victims to further harm or retaliation.
All artificial intelligence systems must comply with South Africa’s Protection of Personal Information Act (POPIA). The playbook provides specific guidance on anonymising survivor data and securing information systems. It also requires that artificial intelligence training data include diverse communities.
South Africa has rural populations, multiple language groups, and various marginalised communities. If artificial intelligence systems only learn from urban, English-speaking data, they will produce biased results. The playbook insists that training data must represent rural women, people with disabilities, and LGBTQIA+ individuals.
The document also recommends creating feedback channels in multiple languages. This allows survivors to report if artificial intelligence systems cause unintended harm or show bias. Regular monitoring can catch problems before they become widespread.
National Disaster Status Brings New Resources to Fight GBV
The November 2025 declaration of gender-based violence as a national disaster changes the legal framework for addressing the crisis. The National Disaster Management Centre classified GBVF under the Disaster Management Act on November 21, 2025. Daily Maverick
This designation potentially unlocks additional resources and coordination mechanisms. Parliament’s oversight committees have stated they will intensify monitoring to ensure the National Strategic Plan moves from paper to practice. Parliament of South Africa
The declaration followed a women’s shutdown on November 21, 2025, where participants held a 15-minute demonstration honouring the estimated 15 women killed daily in South Africa. The New Humanitarian Over one million people across the country participated in the action, which created sustained pressure on the government to respond.
Dr. Makone emphasises that the FemAI Policy Playbook’s launch timing is critical. “As South Africa confronts the newly declared national disaster of GBVF, the FemAI Policy Playbook offers a timely, evidence-based roadmap for using technology, collaboration, and coordinated leadership to drive meaningful change.”
Implementation Requires Long-Term Commitment
The FemAI Policy Playbook provides a roadmap, not a quick fix. Implementing artificial intelligence systems across government departments takes years. It requires sustained political will, consistent funding, ongoing training, and regular evaluation.
Short-term pilot projects often fail because they lack long-term support. The playbook stresses that artificial intelligence solutions need integration into permanent budget processes and operational structures. Technology vendors cannot simply install systems and leave.
Successful implementation also requires cultural change within government departments. Officials must learn to share data across traditional boundaries. They must trust new technologies while maintaining appropriate scepticism. They must prioritise survivor needs over bureaucratic convenience.
The playbook’s official launch on November 19, 2025, presented key findings to government officials, academics, civil society organisations, and the private sector. This broad audience ensures that multiple sectors understand their roles in implementation.
South Africa now has both the policy framework and the crisis declaration to address gender-based violence comprehensively. The FemAI Policy Playbook offers concrete steps for using artificial intelligence as part of this national response. Success depends on translating these recommendations into funded, coordinated action.
Download the policy playbook here.
