Original Summary

Hello HN, I am Kartikeya &amp; we are building the best AI Slop Detector - Checkobot.<p>The current AI Image Detectors are functionally jagged. Some are not able to detect deepfakes, some fail on the latest flux models. Some have an unacceptable false flag rate on genuine images.<p>We have been solving the deepfake detection problem at our parent company - SpoofSense.ai for a while now &amp; we wanted to train a general purpose AI Image Detector that actually works.<p>Checkbot does 3 things: 1.Full Image Model - Detect for AI generator signals in the entire image. 2. Face Specialist - Focuses only on the face to spot face-swaps, edits. 3. C2PA - Checks if the file has signed Content Credentials that record AI generation or editing.<p>Free checks are processed in memory and not stored; we keep the SHA-256 and the verdict. 3 free checks a day.<p>I&#x27;d like to hear where it gets things wrong, especially images it misses or flags unfairly.

  • 情报分类：技术学习与提效
  • 分类依据：内容涉及技术、AI、软件工具或工程实践
  • 信息来源：Hacker News 新项目
  • 发布时间：2026/10/15 03:52:36