Science ❯ Computer Science ❯ Artificial Intelligence

Ethics in AI

Bias in AI AI Safety AI Limitations Fairness in AI AI Research and Development Misinformation Management Behavioral Control Human Impact Future of AI Model Behavior Motivated Reasoning Safety in AI Models Safety and Alignment Value Alignment Social Biases Data Compliance Public Benefit Corporations Risk Assessment Human Values Alignment Safety Concerns Proprietary Technology Misuse of AI AI Development Explainable AI Safety Protocols Reliability in AI Systems Bias Mitigation Privacy Concerns Subliminal Learning Military Applications Model Interpretability

Nature Study Finds ‘Subliminal Learning’ Lets LLMs Inherit Hidden Traits From Synthetic Data

The result spotlights fresh safety risks for distillation, the common practice of training smaller models on outputs from larger ones.

DeepMind Expands Frontier Safety Framework to Target Shutdown Resistance and Harmful Manipulation

Anthropic Unveils Automated Persona Vectors to Steer and Shield Language Models