Spaces:
Running
Running
| title: AI Image Detector | |
| emoji: 🛡️ | |
| colorFrom: blue | |
| colorTo: indigo | |
| sdk: gradio | |
| python_version: "3.12" | |
| app_file: app.py | |
| pinned: true | |
| short_description: Open-baseline AI-generated image detection by Scam.AI | |
| license: apache-2.0 | |
| # AI Image Detector — Scam.AI 🛡️ | |
| Upload any image and an open-source baseline (`Organika/sdxl-detector`) will tell you whether it thinks the image is AI-generated. | |
| **This is a baseline demo**, not a production system. Open detectors are known to fail on: | |
| - Newer generators (GPT-Image-2, FLUX, Imagen 3) | |
| - Post-processed images (super-resolution, JPEG re-compression) | |
| - Out-of-domain content (documents, medical, surveillance) | |
| We published a [comprehensive benchmark](https://huggingface.co/datasets/Scam-AI/gpt-image-2) showing >30 AUC point gaps between academic and real-world performance. | |
| For production-grade detection (calibrated, adversarial-robust, domain-specific) — visit **[scam.ai](https://www.scam.ai)**. | |
| ## Built by Scam.AI | |
| Detection systems for AI-driven fraud. See our [research](https://www.scam.ai/en/research) and [open datasets](https://huggingface.co/Scam-AI). | |