Detecting Real-World AI Scheming with Open-Source Intel
Discover how open-source intelligence uncovers real-world AI scheming incidents, enhancing safety and policy development in AI systems.
CLIP-Inspector: Detect Backdoors in Prompt-Tuned CLIP Models
Detect and repair backdoors in prompt-tuned CLIP models using CLIP-Inspector's OOD trigger inversion for enhanced ML model security.
DeepGuard: Multi-Layer Secure Code Generation with LLMs
DeepGuard enhances secure code generation by aggregating multi-layer semantic cues in LLMs, improving vulnerability detection and preserving code correctne...
Integrating Graph Embeddings in Event Sequence Models
Boost event sequence model accuracy by integrating graph-based embeddings for improved fraud detection and recommendations in digital platforms.
NyayaMind: AI for Transparent Legal Reasoning in India
NyayaMind enhances Indian legal decisions with AI-driven transparent reasoning and accurate judgment prediction, supporting fair and informed outcomes.