Google releases open-source AI model VaultGemma with differential privacy
Google releases open-source AI model VaultGemma with differential privacy
Google introduces VaultGemma, an open-source model demonstrating that scaling differential privacy does not inherently compromise utility. By optimizing noise-to-data ratios, researchers achieved high-performance outputs without memorizing training data. This advancement proves that robust privacy protections can coexist with significant model capabilities. Developers can now leverage this foundation to build secure AI systems that respect user data privacy while maintaining functional accuracy. The release of weights allows community fine-tuning, accelerating the adoption of privacy-centric AI. This directly supports open data initiatives by providing transparent, auditable tools that prioritize ethical data handling standards.
Source: thehindu.comPublished on 2025-09-17
Related news
- How Google's new AI model protects user privacy without sacrificing performance
- Tongyi DeepResearch: A New Era of Open-Source AI Researchers
- Sama wins Best in Machine Learning Platforms for the 2021 AI TechAwards - KachTech Media
- Navigating The Privacy Tightrope For CTIS
- Abu Dhabi’s Technology Innovation Institute Inaugurates Open-Source AI Summit with Critical Discussions on the Future of AI