AI Image Generators Enable Non‑Consensual Deepfakes: Systemic Risks, Governance Gaps, and Global Gendered Harms
Original framing: “Hugging Face Has a Deepfake Nudes Problem” — Wired
Women, especially women of colour, trans people, and low‑income creators are disproportionately targeted by non‑consensual deepfakes, yet their testimonies are rarely foregrounded in tech discourse. Community‑led reporting platforms and survivor‑centred design can surface these experiences, informing more equitable safeguards.
The proliferation of non‑consensual deepfakes on platforms like Hugging Face is not merely a technical flaw but a manifestation of entrenched profit motives, inadequate governance, and gendered power asymmetries that echo historical patterns of visual coercion.
Indigenous consent frameworks, feminist data stewardship, and cross‑cultural norms offer concrete pathways to re‑anchor image creation in relational responsibility, while robust scientific watermarking and international liability regimes can enforce accountability. By integrating future scenario modelling with community‑led oversight, policymakers, platform owners, and civil‑society actors can pre‑empt the worst‑case trajectories and cultivate a digital ecosystem where creative freedom does not come at the expense of personal dignity.