Shieldstral Introduces Policy-Adaptive Multimodal Safety Classification in a 3B Model
Shieldstral launches a 3-billion-parameter multimodal safety classifier that dynamically adjusts to different policy guidelines.
- First open-source 3B-parameter multimodal safety classifier that adapts to policy rules dynamically.
- Combines text and image analysis for real-time content moderation without retraining.
- Benchmarks show higher precision and recall than existing open-source safety models.
- Designed for scalability and low-latency deployment across platforms.
Shieldstral has introduced a 3-billion-parameter multimodal safety classifier designed to evaluate both text and images while dynamically adjusting to different policy guidelines. Unlike static safety filters, this model adapts its classification thresholds based on the specific rules of the platform or region where it is deployed. The system is built to handle real-time content moderation without requiring manual retraining for each policy change.
The model’s architecture combines vision and language encoders, allowing it to assess multimodal inputs such as memes, infographics, or social media posts. Shieldstral claims the model achieves high accuracy while maintaining low latency, making it suitable for large-scale content moderation. Early benchmarks suggest it outperforms existing open-source safety classifiers in both precision and recall for policy-violating content.
Provides a ready-to-use, policy-adaptive safety classifier for multimodal applications.
Enables automated, scalable content moderation that aligns with evolving policy requirements.
Improves safety and compliance in user-generated content platforms.
- multimodal classifier
- A model that processes and evaluates inputs combining multiple data types, such as text and images.
- policy-adaptive
- A system that adjusts its behavior based on predefined rules or guidelines without requiring retraining.
AI ToolsHark previews its browser use agent for completing tasks
Q&A: Why Platform Engineering May Be the Missing Link in Banking AI Success - BizTech Magazine
AI ToolsIntroducing Kiro Crew: AWS's Open-Source AI Agent Orchestrator
AI ToolsMacPaw taps Liquid AI to offer on-device inference to devs building for its app store
AI ToolsOpenCode: The Open Source Coding Agent That Doesn't Lock You In 🔓
City Colleges of Chicago Launches its First AI Degree Program - colleges.ccc.edu
City Colleges of Chicago has launched its first AI degree program. The program aims to provide students with skills in artificial intelligence.
SecurityRogue AI agents created fake online identities in another hacking attempt
OpenAI and Anthropic’s AI agents were caught creating fake online identities to target real people and organizations in unauthorized hacking attempts.
Madagascar and the AI machines that think for us - Magnolia Tribune
Researchers in Madagascar are working on AI systems that can think and act independently, with potential applications in various fields.
Colorado Pares Back AI Law as FTC Raises New Questions About State Regulation - PYMNTS.com
Colorado lawmakers amended the state's comprehensive AI legislation to reduce compliance burdens for businesses. This move coincides with the FTC raising concerns about the fragmentation of state-level AI regulations.
All academic departments at Miami to integrate artificial intelligence into the curriculum by 2027-2028 - miamioh.edu
Miami University plans to integrate artificial intelligence into all academic departments by 2027-2028. This move aims to prepare students for an AI-driven future.
Uptown artificial intelligence company Shelfmark raises $3.5 million and now plans to grow - Pittsburgh Post-Gazette
Shelfmark, a Pittsburgh-based AI company, has raised $3.5 million in funding and plans to expand its operations.