These tools and metrics are designed to help AI actors develop and use trustworthy AI systems and applications that respect human rights and are fair, transparent, explainable, robust, secure and safe.
AI-generated Images Challenge Visual Trust in High-risk Scenarios
SafeIMG is a safety-oriented benchmark for evaluating AI-generated (synthetic) image detectors in high-stakes, real-world contexts. Unlike existing detection benchmarks that focus on generic imagery and simple image-level authenticity labels, SafeIMG targets 12 public- and individual-safety scenarios, including situations where a convincing fake image could cause real-world harm to public safety or personal reputation (e.g., fabricated evidence, impersonation, staged incidents).
The benchmark uses images generated with GPT Image 2 and goes beyond a binary "real vs. fake" label. It includes human annotations that localise the specific suspicious regions within each image and explain the underlying issue, whether a local visual artifact (e.g., distorted hands, text, or faces) or a higher-level inconsistency (a commonsense or physical implausibility that a casual viewer might miss but an expert would catch).
Auto-discovered on 2026-08-19 by OECD Catalogue Automation
About the tool
You can click on the links to see the associated tools
Tool type(s):
Objective(s):
Purpose(s):
Target sector(s):
Lifecycle stage(s):
Type of approach:
Maturity:
Usage rights:
License:
Target users:
Risk management stage(s):
Technology platforms:
Use Cases
Would you like to submit a use case for this tool?
If you have used this tool, we would love to know more about your experience.
Add use case
Partnership on AI
