The rise of toxic content on digital platforms has intensified the demand for automatic moderation tools. While English has benefited from large-scale annotated corpora, Spanish remains under-resourced, particularly for nuanced cases of toxicity such as irony, sarcasm, or indirect aggression. We present an extended version of the NECOS-TOX corpus, comprising 4,011 Spanish comments collected from 16 major news outlets. Each comment is annotated across three levels of toxicity (Non-Toxic, Slightly Toxic, and Toxic), following an iterative annotation protocol that achieved substantial inter-annotator agreement (k = 0.74). To reduce annotation costs while maintaining quality, we employed a human-in-the-loop active learning strategy, with manual correction of model pre-labels. We benchmarked the dataset with traditional machine learning (ML) methods, domain-specific transformers, and instruction-tuned large language models (LLMs). Results show that compact encoder models (e.g., RoBERTa-base-bne, 125M parameters) perform on par with much larger models (e.g., LLaMA-3.1-8B), underscoring the value of in-domain adaptation over raw scale. Our error analysis highlights persistent challenges in distinguishing subtle forms of toxicity, especially sarcasm and implicit insults, and reveals entity-related biases that motivate anonymization strategies. The dataset and trained models are released publicly.

The current best practice to measure the performance of base Large Language Models is to establish a multi-task benchmark that covers a range of capabilities of interest. Currently, however, such benchmarks are only available in a few high-resource languages. To address this situation, we present IberoBench, a multilingual, multi-task benchmark for Iberian languages (i.e., Basque, Catalan, Galician, European Spanish and European Portuguese) built on the LM Evaluation Harness framework. The benchmark consists of 62 tasks divided into 179 subtasks. We evaluate 33 existing LLMs on IberoBench on 0- and 5-shot settings. We also explore the issues we encounter when working with the Harness and our approach to solving them to ensure high-quality evaluation.

Recent advancements in Natural Language Processing (NLP) have allowed systems to address complex tasks involving cultural knowledge, multi-step reasoning, and inference. While significant progress has been made in text summarization guided by specific instructions or stylistic cues, the integration of pragmatic aspects like communicative intentions remains underexplored, particularly in non-English languages. This study emphasizes communicative intentions as central to summary generation, classifying Spanish product reviews by intent and using prompt engineering to produce intention-aligned summaries. Results indicate challenges for large language models (LLMs) in processing extensive document clusters, with summarization accuracy heavily dependent on prior model exposure to similar intentions. Common intentions such as complimenting and criticizing are reliably handled, whereas less frequent ones like promising or questioning pose greater difficulties. These findings suggest that integrating communicative intentions into summarization tasks can significantly enhance summary relevance and clarity, thereby improving user experience in product review analysis.

Fever Shared 2.0 Task is a challenge meant for developing automated fact checking systems. Our approach for the Fever 2.0 is based on a previous proposal developed by Team Athene UKP TU Darmstadt. Our proposal modifies the sentence retrieval phase, using statement extraction and representation in the form of triplets (subject, object, action). Triplets are extracted from the claim and compare to triplets extracted from Wikipedia articles using semantic similarity. Our results are satisfactory but there is room for improvement.