gemma-tuner-multimodal

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.90). The skill explicitly supports streaming and ingesting datasets from external GCS/BigQuery buckets ("GCS Streaming for Large Datasets" and the [dataset:large-audio-gcs] example) and loading checkpoints from Hugging Face (e.g., base_model / HF auth), meaning it will consume arbitrary public/user-uploaded content that can materially change training behavior and downstream actions.