Sentence-level AI-generated text detection (S-AGTD) for hybrid documents, where humans and LLMs co-author one text, faces two gaps: existing methods classify each sentence in isolation, discarding inter-sentence dependencies, and existing benchmarks omit the newest generation of generators. We construct MOSAIC, a benchmark of 16,000 hybrid documents over PubMed and XSum, generated by DeepSeek-V3.
arXiv:2606. 06481v1 Announce Type: cross Abstract: As AI writing assistants become increasingly integrated into real-world drafting and revision workflows, many documents are no longer purely human-written or AI-generated, but instead result from progressive human-AI co-editing.
The paper introduces Relational Over-Regularization (ROR), a new way to detect AI‑generated text by examining sentence‑pair transition variance rather than token‑level cues. It proposes the Cross‑Source Stylometric Fingerprint Graph (CSFG), a graph‑based model that encodes positional, sequential, semantic, and transition deviation signals as learnable edge features, achieving 97.14% accuracy on binary detection and outperforming existing graph baselines by 11.14 percentage points. The approach shows strong generalization to unseen large language models in the inflated‑variance regime while maintaining a low false‑positive rate of 1.57%.
Distinguishing machine-generated text (MGT) from human-written text (HWT) becomes increasingly important due to potential misuse. However, most supervised detectors often degrade out-of-domain (OOD) a...
The paper introduces SCSP, a training‑free framework that improves long‑context embeddings by selectively pooling informative tokens. SCSP partitions documents into sentence‑aware chunks, adds a semantic compression prompt to each chunk, and uses prompt‑isolated attention masks to estimate token importance. The selected tokens’ intermediate‑layer representations are aggregated to form the final embedding, yielding consistent performance gains across zero‑shot and fine‑tuned models on long‑context benchmarks.
arXiv:2605. 00924v2 Announce Type: replace-cross Abstract: AI-generated content (AIGC) detectors are increasingly deployed in high-stakes settings such as academic integrity screening, yet their reliability rests on a fundamental paradox: as language models are trained on human-written corpora, the statistical boundary between AI and human writing will inevitably dissolve as models improve.
The paper introduces Relational Over‑Regularization (ROR), a structural signal at the sentence‑pair level that captures inflated inter‑sentence transition variance in AI‑generated text. It proposes the Cross‑Source Stylometric Fingerprint Graph (CSFG), a graph‑based framework that encodes positional, sequential, semantic, and transition deviation signals as learnable GNN edge features, achieving 97.14% accuracy in binary detection and outperforming existing graph‑based baselines. The method demonstrates robust generalization to unseen large language models in the inflated‑variance regime while maintaining a low false‑positive rate.
arXiv:2607. 03680v1 Announce Type: new Abstract: Recent AI-generated text detection work often introduces a new benchmark together with a specialized detector tailored to it.
arXiv:2607. 17524v1 Announce Type: cross Abstract: We propose Token-Level Off-Policy Labeling (TOPL), an off-policy training paradigm that reframes post-training as a token-level correctness prediction task.
arXiv:2609.13481v1 Announce Type: new Abstract: Large language models have made abstractive summarization remarkably fluent, but generated summaries can hallucinate facts, posing serious risks in bio...
arXiv:2609.14352v1 Announce Type: new Abstract: AI-generated image detection has attracted increasing attention, but existing evaluations mainly focus on natural images, leaving AI-generated document...
We propose Token-Level Off-Policy Labeling (TOPL), an off-policy training paradigm that reframes post-training as a token-level correctness prediction task. Our key intuition is that by training the model to distinguish good and bad tokens in a response, we naturally guide the model towards generating good tokens, while avoiding the pitfalls that come with directly training the model to generate off-policy tokens.
arXiv:2606. 04442v1 Announce Type: cross Abstract: AI systems increasingly need to combine two demanding capabilities: navigating multi-session conversation history and performing deep reading comprehension within long documents.