A Unified Detection Framework for AI-Related Content and Artifacts

2026-07-08

AI Summary

Existing approaches struggle to uniformly and efficiently detect diverse forms of AI-generated content, including large language model outputs, hallucinations, watermarked texts, and adversarial examples. This work proposes a unified detection framework based on Mahalanobis distance scoring, which leverages deep representation learning to accurately model the covariance structure of in-distribution (positive) samples—such as human-written text or factual statements. The method innovatively introduces a joint Minimum Covariance Determinant (MCD) estimator tailored for multi-class positive data, offering both high breakdown-point robustness and theoretical convergence guarantees. Extensive experiments demonstrate that the proposed framework achieves strong robustness and broad applicability across a variety of AI-generated content detection tasks.

Abstract

Artificial intelligence (AI) is a double-edged sword: while it has achieved remarkable success across a wide range of domains, its deployment also calls for effective oversight and regulation, for which the detection of AI-related content and artifacts is perhaps the most direct and cost-effective approach. To this end, we propose a unified detection framework based on Mahalanobis distance scores (MDS), applicable to several important settings, including the detection of large language model (LLM) generated text, hallucination, watermark, and adversarial examples. A key component of the proposed method is to accurately characterize the positive class--such as human-generated text, factual statements, unwatermarked text, or non-adversarial samples--which requires an efficient and robust estimator of the covariance matrix of deep representations of positive samples before computing the MDS. Since the positive samples typically consist of multiple classes, and these classes may exhibit both homogeneity and heterogeneity, we develop joint estimation methods for both the casewise and cellwise minimum covariance determinant (MCD) estimators. We provide efficient optimization algorithms for both estimators and prove their convergence. We provide a reasonable definition of the breakdown point for the joint estimators and prove their corresponding high breakdown point properties. Empirical evaluations confirm the effectiveness of the proposed detection framework.

Problem

• AI-generated content detection
• hallucination detection
• watermark detection
• adversarial example detection
• AI oversight

Innovation

• Mahalanobis distance
• minimum covariance determinant
• unified detection framework
• robust covariance estimation
• AI-generated content detection