is the third iteration of the GAIA (Global Abuse Identification and Analytics) series, a deep‑learning system aimed at detecting and flagging visual content that depicts or encourages facial abuse (e.g., non‑consensual deepfakes, facial manipulation for harassment, or exploitative imagery).
| Strengths | Limitations | |-----------|-------------| | • State‑of‑the‑art detection performance (AUROC ≥ 0.94).• Multimodal (image + short video) support.• Prompt‑based zero‑shot adaptability.• Open‑source, well‑documented code and model card.• On‑device inference option for privacy. | • Large model size; heavy compute for real‑time video.• Temporal window limited to ≤ 30 s.• Slight bias in certain sub‑categories (e.g., forced distortion).• Explanations sometimes generic, not always actionable.• No built‑in adversarial robustness against targeted evasion. | Facialabuse-gaia-3