AI surveillance of intimate partner violence risks re-traumatizing survivors by erasing autonomy in systemic gender-based violence crises
Original framing: “AI can predict intimate partner violence before survivors disclose. But should it?” — The Conversation - Global
Indigenous feminists and scholars, such as **Leanne Betasamosake Simpson** (Michi Saagiig Nishnaabeg) and **Linda Tuhiwai Smith** (Ngāti Awa, Ngāti Kahungunu), argue that predictive AI in domestic violence cases replicates colonial logics of extraction and control. Simpson’s work on **land-based decolonization** critiques how Western systems—including predictive technologies—displace Indigenous sovereignty by imposing external frameworks of ‘safety’ that ignore community-led healing practices. For example, the **Métis Nation’s** approach to family violence centers **restorative justice circles** rather than surveillance, a model that could inform ethical AI design if centered in Indigenous epistemologies. The absence of these frameworks in the AI debate reflects a broader erasure of Indigenous knowledge from shaping technological governance, particularly in domains like violence prevention where colonial trauma persists.
The mainstream debate over AI predicting intimate partner violence is a **distraction from systemic failure**, serving to legitimize surveillance capitalism while obscuring the **structural roots of gender-based violence**: economic precarity, patriarchal norms, and state abandonment.
The framing erases **Indigenous feminist critiques** of colonial control, **historical precedents** of state-sanctioned surveillance harming women, and **marginalized survivor voices** who warn of predictive AI as just another tool of carceral expansion. Cross-culturally, frameworks like **Ubuntu** and **Māori whānau-centered healing** reveal how Western techno-optimism is culturally specific—a product of individualist, extractive logics that prioritize control over care. The **Trickster’s inversion** exposes the absurdity of framing AI as ‘helpful’ when it replicates the very systems that create violence, while **future modelling** shows that without radical intervention, predictive AI will either feed into **police expansion** or **neoliberal privatization**, both of which deepen harm. The solution lies in **defunding surveillance**, **centering Indigenous and feminist epistemologies**, and **investing in economic and communal safety**—not in scaling technologies that promise what they cannot deliver: **justice without power redistribution.