Latest Articles for Safety

Robots are changing how we live and work in various settings.

A formal approach to verify the correctness of DAG-based consensus protocols.

Research shows simple input changes can lead to harmful outputs in LLMs.

A new model improves safety monitoring for large language models against harmful content.

RSHPN framework enhances reliability and safety in robotic systems through better analysis.

Combining Hamilton-Jacobi reachability with reinforcement learning enhances safety in control systems.

A new benchmark aims to assess AI safety risks effectively.

Regulations guide the safe and fair use of AI technologies across various sectors.

A look into bias in language models and their impact on fairness.

Logic provides a strong foundation for effective learning in science and engineering.

Examining the risks and safety measures for LLM-powered robots.

Humanoid robots are changing how we approach work in various industries.

A look into assessing mental wellness chatbots for effective user support.

Phi-3 models focus on safety and aligning with human values.

Examining how explainable AI can aid in certifying AI systems.

A look at AI's potential to improve tractor safety.

This paper outlines guardrail models to promote safe AI in education.

Exploring machine unlearning and its role in enhancing generative AI safety and privacy.

A program to teach model checking for reliable software development.

A new framework improves safety in self-driving cars through formal methods.

A new framework improves safety for AI-controlled cyber-physical systems.

Examining strategies to create AI systems that prioritize safety and minimize risk.

Gemma Scope offers tools for better understanding language models and improving AI safety.

Explore the role of Control Barrier Functions in maintaining safety across various systems.

A new system enhances vehicle control in rough terrain using two methods.

This study examines injuries resulting from robot-human interactions in industrial settings.

A new approach to enhance how robots understand and respond to users.

A novel approach streamlines control design while ensuring safety in nonlinear systems.

CERN upgrades beam stoppers to boost safety and efficiency for high-energy particle beams.

Clarifying operational limits ensures safety for self-driving cars.

Caution-Aware Transfer enhances safety and performance in reinforcement learning applications.

A novel approach to self-driving vehicles that prioritizes safety and comfort.

A new method improves the safety verification of neural networks in critical systems.

A guide to managing diverse objectives in decision processes using aspiration sets.

A new approach streamlines safety and helpfulness in language model training.

Researching new treatments through small experiments prioritizes patient safety.

A comprehensive evaluation framework for healthcare chatbots is introduced to enhance their effectiveness.

Examining the effects of updates on safety, bias, and authenticity in image generation.

A novel approach improves drifting safety for self-driving cars using physics.

A novel approach to optimize points-to analysis, balancing speed and accuracy.