Watermarking has become a key mechanism for distinguishing AI-generated from human-written texts. In practical settings, however, content produced by large language models (LLMs) is often subject to critical post-generation edits, posing challenges for detecting and localizing such modifications. In this work, we propose incorporating error-correcting codes, specifically Varshamov-Tenengolts (VT) codes, by embedding VT code strings into watermarked LLM outputs, thus enabling post-generation edit detection of token insertions and deletions. We accompany such a watermarking procedure with a watermark detector and a post-edit detector. We evaluate our method on open-source LLM using synthetic edits and demonstrate excellent empirical performance of post-generation edit detection.

We describe the Qahera system for the four IslamicEval 2026 subtasks: (1) detecting Qur’an/Hadith citation spans in LLM-generated Islamic answers, (2) judging citation authenticity, (3) correcting inauthentic citations against reference corpora, and (4) judging whether an authentic citation answers the question asked. Under a 13B-parameter open-model constraint, running entirely on free Kaggle GPUs, we combine a lightweight token-classification model with character-level auxiliary features (Sub-task 1), corpus-grounded fuzzy retrieval (Sub-tasks 2–3), and a 4B-parameter instruction-tuned LLM (Gemma-3-4B) used only where retrieval alone cannot discriminate (Subtasks 3–4). Rather than hand-designing LLM prompts once, we develop them through an iterative propose-evaluate-diagnose search loop that reads full prediction traces after every candidate and hunts for systematic biases before each change. This surfaces small-model failure modes – a near-universal “reject” bias in isolated yes/no verification, and an “always name something” bias in numbered-list prompts – that would otherwise silently cap performance, and shows that validating candidates on small subsets can be actively misleading. Our final system reaches 0.9716 F1 (Subtask 1), 0.906 accuracy (Subtask 2), 0.773 accuracy (Subtask 3), and 0.673 macro-F1 (Subtask 4) on the official test sets.

We submit systems to both the closed and open divisions of all four tracks of the AraSeg 2026 shared task (PA, NP, NoPnx-PA, NoPnx-NP): a supervised 3-backbone ensemble (CAMeLBERT-MSA, AraBERTv02, and AraELECTRA, each with a gated-residual head, R-Drop regularization, and stochastic weight averaging), and a zero-shot LLM system prompted with explicit, track-specific segmentation rules instead of labeled examples. The ensemble uses no external resources, so it is entered in both divisions everywhere; for NoPnx-NP specifically, where every structural cue is stripped away, we additionally blend its predictions with the LLM system’s, in a real-data-only variant (closed-eligible) and a wiki-augmented variant trained with added pseudo-labeled Wikipedia data (open-only). On the officially-scored blind test, our best submission per track reaches 94.70 F₁ (PA), 91.00 (NP), 91.20 (NoPnx-PA), and 87.40 (NoPnx-NP). The supervised ensemble is the stronger, more consistent choice overall; the one exception is NoPnx-PA, where our LLM system’s unmodified seed prompt outperforms it and combining the two systems does not help—the only track where that happens. We report in full the structural annotation ambiguities we uncovered while building the LLM system.