Agentic ai improves open world fake image detection by better evidence handling
Dissecting Agentic Forensics: The Role of Triage, Prompting, and Evidence Arbitration in Open-World Fake Image Detection
Computer Vision and Pattern RecognitionArtificial IntelligenceCryptography and Security
Summary
Detecting fake or edited images is tricky because many tools only look for specific types of changes. The authors study a system that uses several specialist detectors and an AI 'judge' to decide which parts of the evidence to trust and how to resolve conflicts. They found that simply combining many detectors leads to many false alarms on real images. The key to better detection is the AI’s reasoning ability to filter unreliable clues and decide which detector is more trustworthy, especially when facing new kinds of image forgeries.
What this means in practice
- •For social media platform security teams: Improve detection of diverse image manipulations by integrating specialist forensic tools with reasoning-based arbitration to reduce false alarms on genuine images.
- •For digital forensics analysts: Enhance accuracy in identifying image tampering by relying on AI-driven evidence filtering and conflict resolution among multiple tools.
Authors
Xianlong Li, Pietro Bongini, Niccoló Pancino, Marco Blanchini, Benedetta Tondi, Mauro Barni
Abstract
Image forensics is increasingly an open-world problem: manipulations range from fully synthetic images to localized edits, splicing and swapping, while most forensic detectors remain specialized to a single manipulation family. Agentic AI has recently emerged as a promising solution. In principle, such systems can assess the reliability of individual detectors, identify out-of-scope evidence, and arbitrate conflicting reports. However, it remains unclear which components actually drive performance and whether their benefits persist under distribution shift. To answer these questions, we study a training-free agentic framework built around specialist detectors, per-detector triage, and conflict-aware evidence arbitration. Using six configurations and three multimodal large language model backbones, we dissect the role of triage, prompting, and reasoning quality on both in-distribution and out-of-distribution data. Our results show that naive detector fusion suffers from severe false-positive rates on authentic images. Triage and prompting consistently improve performance by filtering unreliable evidence and exposing detector limitations. However, the dominant factor is represented by reasoning itself: A stronger judge substantially outperforms a weaker one, particularly under distribution shift. Most notably, manipulation recall is nearly saturated across all configurations, indicating that the main challenge of open-world image forensics is not detecting manipulations, but calibrating trust in specialized forensic tools and arbitrating conflicting evidence.