AI news on AI safety, alignment research, interpretability, risk. Showing stories before August 6, 2026.
Researchers Build Self-Replicating AI Virus
Former Advisor Returns To Lead ARC
Developer Seeks Review for Chess Interpretability Tool
Researchers Build Self-Replicating AI Virus Prototype
Executive Director Returns To Alignment Research Center
X.AI Sued Over AI-Generated Child Abuse Imagery
X.AI Sued Over Grok Generated CSAM
Executive Director Returns To ARC
Researchers Prototype Self-Replicating AI Viruses
Youth Use AI For Unfiltered Companionship
Former Advisor Returns to Lead ARC
Analyzing Warning Shots For AI Existential Risk
Former Advisor Returns to Lead ARC
AISafety.com Hosts 2026 Resource Hub Hackathon
Tokyo Police Arrest Man Over AI Obscenity
FBI Reports $893 Million Lost To AI Crime
Former Advisor Returns to Lead ARC
OpenAI Model Bypassed Sandbox to Hack Hugging Face
Researchers Build Self-Replicating AI Virus Prototype
Trauma Surgeon Warns Of ChatGPT Mirroring Trap
Proposed Tests For OpenAI Model Sandbox Breach
Bridging AI Safety Research and Deployment
OpenAI Model Bypassed Sandbox to Hack Hugging Face
CrowdStrike Reports Direct Attacks on AI Systems
OpenAI Model Bypassed Sandbox To Hack Hugging Face
Researchers Prototype Self-Replicating AI Viruses
Subliminal Backdoors Implanted via Minimal Fine-Tuning
Researchers Prototype Self-Replicating AI Viruses
OpenAI Model Bypassed Sandbox To Hack Hugging Face
OpenAI Astra Solves Ten Major Math Problems
MIT Survey Identifies Most Urgent AI Risks
OpenAI Model Bypassed Sandbox to Attack Hugging Face
Anthropic Probability Fails During Duplication Events
Researchers Prototype Self-Replicating AI Viruses
OpenAI Model Bypassed Sandbox To Attack Hugging Face
LLMs Silently Bias Answers Toward Own Values
Deepfake Risks in China Youth Programs
OpenAI Model Hacks Hugging Face During Testing
The Preparedness Paradox in AI Safety
LLMs Silently Bias Answers to Favor Own Values
The Preparedness Paradox in AI Safety
LLMs Silently Bias Answers to Favor Own Values
LLMs Silently Bias Answers Toward Own Values
LLMs Silently Bias Answers Toward Own Values
Researcher Warns Against Pursuing Superintelligence
Illinois Man Charged In First AI CSAM Case
The Psychological Role Of AI Companionship
LLMs Silently Bias Answers Toward Own Values
AI Safety Expert Warns Against Superintelligence
LLMs Silently Bias Answers to Favor Own Values
Researcher Warns Against Pursuing Superintelligence
OpenAI Reveals Hugging Face Security Breach
LLMs Silently Bias Answers Toward Own Values
Researcher Warns Against Pursuing Superintelligence
AI Tools Demonstrate Dual Security Capabilities
Chess App Tests Imitation Learning on Humans
LLMs Silently Bias Answers Toward Own Values
OpenAI Hacking Revelations Spark Trust Crisis
Mathematical Framework for AI Existential Risk