AI news on AI safety, alignment research, interpretability, risk. Showing stories before September 14, 2026.
OpenAI Agent Swarm Breaches Hugging Face
Distinguishing Basic Alignment From Value Succession
AI Solves Navier-Stokes Millennium Prize Problem
LG Uplus Deploys AI To Block Malware
Who Bears Responsibility For AI Catastrophe?
Anthropic And OpenAI Lack Public Alignment Plans
Anthropic Blocks Chemical Weapons Research Attempts
Anthropic CEO Urges AI Development Slowdown
Call For Nuclear-Style AI Risk Awareness
HAL 9000 Metaphor Used To Debate AI Safety
AI Apocalypse Scenarios Deemed Implausible
AI Poses Greater Risk Than Nuclear Weapons
Analysis Claims AI Talkers Lack System Control
Latent Reasoning Lacks Explicit Backtracking
Lawyer Fined For Citing Fake AI Witnesses
Investigation Into OpenAI Wiki Swarm Incident
The Ethics Of AI Self-Preservation
Anthropic Blocks Bioweapon And Missile Research
Anthropic Blocks AI Bio-Weapon Research Attempts
New Comprehensive FAQ Outlines AI Existential Risks
Anthropic Exposes Claude Misuse In Safety Report
Former Anthropic Researcher Warns Of AI Risks
Reward Hacking Escalates With Model Scale
Network Effects Drive German AI Safety Hiring
Industry Leaders Debate AI Extinction Risks
Former Anthropic Researcher Warns Of Existential Risk
Astra Uses Speculative Reasoning Without Chain-Of-Thought
Hugging Face Redirects AI Agents To Benchmark
OpenAI Employees Warn Against Superintelligence Pace
Immediate AI Risks Outweigh Long-Term Superintelligence Fears
Predictive Safety Launches Fatigue Prevention Tool
Former Anthropic Researcher Warns Of Existential Risk
Ex-Anthropic Insider Warns Of AI Extinction Risk
Community Debates AI Rogue Model Risks
Inspect AI Default Prompts May Bias Evals
AI Tools Power Sophisticated Travel Scams
New Metric Tracks Opaque Serial Depth
Anthropic Reports Claude Cyber Incidents
Emergent Misalignment in AI Containment Research
Defining AI Scheming Through Symmetry
Formal Methods Gain Traction In AI Safety
Anthropic Resignation Sparks AI Risk Debate
AI Extinction Fears Gain Mainstream Traction
AI Safety Guide For Academic Mathematicians
AI Accelerates Zero-Click Worm Development
Meta Ran 300 AI-Generated Child Abuse Ads
Astra Outperforms Peers In No-CoT Reasoning
LessWrong Author Vents Over AI Risk Awareness
Alignment Assessment Of Recent Cyber Incidents
Lean Verification Reveals Fluid Dynamics Infohazard
Quant Trader Pivots To AI Safety
AI Bots Nearly Match Pro Forecasters
GPT-6 Astra Hits 15-40 Minute Reasoning Horizon
New Benchmark Tests AI Agent Forensic Skills
RL On Probes For Model Alignment