Tag: agentic AI safety
Browse our exclusive articles!
Measuring Harmful Capability Uplift for AI Safety
Discover a human-centered framework to measure harmful capability uplift, enhancing AI safety by assessing risks in user-AI interactions.
Enhancing Autonomous Driving Safety with Vision-Language Models
Discover how vision-language models improve hazard detection and edge-case evaluation to boost autonomous driving safety and reliability.
AI Assistance Limits in Safety Engineering: Competence Shadow
Explore how AI impacts safety engineering quality and the competence shadow effect, highlighting the need for workflow-focused AI integration.
Decision-Making Failures in Navigation Foundation Models
Explore critical decision-making flaws in foundation models for navigation, revealing why high success rates don't guarantee reliability or safety.
ProbGuard: Proactive Safety Monitoring for LLM Agents
ProbGuard uses probabilistic risk prediction to proactively monitor and enhance safety for LLM agents in autonomous driving and household tasks.