This document proposes a weakly supervised method to extract causality knowledge from Wikipedia articles in multiple languages. It collects causality seed entities and contexts using predefined keywords and entity links between articles. A classifier is trained on these distant examples to predict causal relations. In experiments, the proposed method achieves over 98% precision and 64% recall on a Wikidata test set, outperforming baselines. Analysis finds most predictions are verifiable through article evidence, demonstrating the method extracts meaningful causal knowledge.