Every information ecosystem produces beliefs that shape strategic decisions. Both human analysts and AI systems inherit the blind spots of their information sources. We show that LLMs, combined with prediction markets, function as a calibrated instrument for measuring how far ecosystem-induced beliefs fall from reality: LLMs extract the beliefs a text corpus implies, and prediction markets provide a ground truth proxy against which to quantify the error.We isolate the bias contribution of specific text through ablation: varying information context while holding the model fixed, with a contaminated model that knows actual outcomes as control. Applied to 111 Ukraine-related prediction markets (~93,000 predictions, four models), we find that English news context systematically biases territorial predictions, wrong 64–72% of the time (p 10{-6}). A contaminated model that knows actual outcomes shows the same error rate, indicating the bias originates primarily in the text. Supplementing with Ukrainian military-analytical sources partially corrects the distortion.We show that the distortion originates primarily in the sources, not the models. Consistent across four architectures, it will persist in any system that processes them and propagate into downstream decisions.

In this paper, we present our solutions for the two UNLP 2025 shared tasks: manipulation span detection and manipulation technique classification in Ukraine-related media content sourced from Telegram channels. We experimented with fine-tuning large language models (LLMs) with up to 12 billion parameters, including both encoder- and decoder-based architectures. Our experiments identified Gemma 3 12b with a custom classification head as the best-performing model for both tasks. To address the limited size of the original training dataset, we generated 50k synthetic samples and marked up an additional 400k media entries containing manipulative content.