The malicious dissemination of hate speech via compromised accounts, automated bot networks and malware-driven social media campaigns has become a growing cybersecurity concern. Automatically detecting such content in Spanish is challenging due to linguistic complexity and the scarcity of annotated resources. In this paper, we compare two predominant AI-based approaches for the forensic detection of malicious hate speech: (1) fine-tuning encoder-only models that have been trained in Spanish and (2) In-Context Learning techniques (Zero- and Few-Shot Learning) with large-scale language models. Our approach goes beyond binary classification, proposing a comprehensive, multidimensional evaluation that labels each text by: (1) type of speech, (2) recipient, (3) level of intensity (ordinal) and (4) targeted group (multi-label). Performance is evaluated using an annotated Spanish corpus, standard metrics such as precision, recall and F1-score and stability-oriented metrics to evaluate the stability of the transition from zero-shot to few-shot prompting (Zero-to-Few Shot Retention and Zero-to-Few Shot Gain) are applied. The results indicate that fine-tuned encoder-only models (notably MarIA and BETO variants) consistently deliver the strongest and most reliable performance: in our experiments their macro F1-scores lie roughly in the range of approximately 46%–66% depending on the task. Zero-shot approaches are much less stable and typically yield substantially lower performance (observed F1-scores range approximately 0%–39%), often producing invalid outputs in practice. Few-shot prompting (e.g., Qwen 3 8B, Mistral 7B) generally improves stability and recall relative to pure zero-shot, bringing F1-scores into a moderate range of approximately 20%–51% but still falling short of fully fine-tuned models. These findings highlight the importance of supervised adaptation and discuss the potential of both paradigms as components in AI-powered cybersecurity and malware forensics systems designed to identify and mitigate coordinated online hate campaigns.

The digitization of healthcare has resulted in the production of large amounts of structured and unstructured clinical data, creating the need for accurate and efficient named entity recognition (NER) to support medical procedures. This study evaluates and compares three approaches to NER in the medical domain in Spanish: using Large Language Models (LLMs) with In-Context Learning techniques (Zero-Shot, Few-Shot, and Chain-of-Thought); fine-tuning of LLMs; and fine-tuning of encoder-only models. Experiments were conducted on the Meddocan, Meddoprof, Meddoplace and Symptemist benchmark datasets. Fine-tuned encoder-only models achieve the best performance across all datasets, reaching macro-F1 scores of up to 76.71 on Meddocan, 71.51 on Meddoplace, 66.07 on Meddoprof and 63.50 on Symptemist. While LLMs with prompting offer flexibility and require no task-specific training, their performance varies significantly depending on the entity type. In addition, we evaluated fine-tuning of LLMs using QLoRA, but the improvements were limited due to the small amount of training data available per entity type, which made model adaptation less effective.