Hey Product Hunt 👋 I’m lucky, the solo maker behind Pixoraid. I started building Pixoraid after noticing a growing problem: AI-generated images are becoming more realistic, but most detection tools still return a single unexplained “real” or “AI” score. That can create false confidence, especially when an image has been compressed, edited, cropped, or screenshotted. Pixoraid takes a more transparent approach. Instead of pretending detection is perfect, it provides an AI-generation probability, confidence level, metadata findings, supporting visual signals, and an “inconclusive” result when the evidence is not strong enough. While building it, my approach evolved from creating a simple binary classifier into developing an evidence-first image analysis product. I learned that real-world testing matters more than clean benchmark accuracy, so I’m focusing on difficult cases such as social-media compression, AI-edited images, realistic phone-style generations, and genuine photos that may look synthetic. Pixoraid is still early, and I’m launching today to learn from real users—not to claim that the problem is completely solved. I’d especially value feedback on: • Whether the result explanation is understandable • Which evidence or signals you would like to see • Images where Pixoraid produces the wrong or uncertain result • Whether an API or batch-analysis workflow would be useful to you Please try it with both real and AI-generated images and share the difficult cases. Every honest test will help me improve the product. Thanks for checking out Pixoraid 💙

Pixoraid

Detect AI-generated images with evidence, not guesses

Detect AI-generated images with evidence, not guesses