Protecting multimodal large language models against misleading visualizations

Connections are not prepared for this record yet, or its source metadata has changed. The original record remains available while background snapshots are rebuilt.