TruthLens introduces a novel training-free framework for detecting AI-generated fake images while providing clear interpretability of results.

• Eliminates dependency on supervised training data, unlike traditional binary classifiers
• Provides visual explanations of why an image is classified as fake
• Utilizes semantic understanding to identify inconsistencies in synthetic content
• Offers robust detection across multiple generative AI models

Security Impact: As sophisticated deepfakes threaten digital trust, TruthLens enables more reliable visual disinformation detection, helping protect against fraud, misinformation campaigns, and identity theft in an increasingly AI-generated visual landscape.

TruthLens: A Training-Free Paradigm for DeepFake Detection