Safer and Multimodal: Responsible AI with Gemma- Google Developers Blog

Introduces ShieldGemma 2, a 4B parameter safety classifier model built on Gemma 3, designed to detect harmful content in images (sexually explicit, dangerous, violence). It can be used as an input filter for vision language models or an output filter for image generation models. ShieldGemma 2 is trained on curated datasets of natural and synthetic images and instruction-tuned Gemma 3. It offers flexibility for fine-tuning, versatility with Gemma 3-supporting frameworks, and is open for community collaboration. Performance benchmarks show optimal F1 scores on internal benchmarks.