Nearly half of all respondents (47%) are seeing a rise in attacks targeting their organization’s large language model (LLM) deployments.