Science ❯ Computer Science

Artificial Intelligence Research

AI Safety Model Development Model Training Model Evaluation AI Models Reinforcement Learning Language Processing Model Capabilities AI Applications Key Figures in AI Machine Learning Peer Review in AI AGI Model Vulnerabilities Benchmarking AI Models General Intelligence Autonomous Systems Advanced AI Models AI Development Human-Computer Interaction Telecommunications Research Model Training Techniques Reasoning Capabilities Natural Language Processing AI Techniques Open Science Model Behavior Analysis

Anthropic-Led Study Finds 250 Poisoned Files Can Backdoor Large AI Models

Researchers say the fixed-count threat makes rigorous data provenance urgent.

Anthropic Unveils Automated Persona Vectors to Steer and Shield Language Models