AI news on AI safety, alignment research, interpretability, risk. Showing stories before September 1, 2026.
Debating The Nature Of AI Catastrophes
New Framework For AI Value Generalisation
Call For Radical AI Safety Funding
New Framework For AI Value Generalisation
Musk And Harari Debate AI Control
New Theory Links Alignment Failures To Value Generalisation
New Study Group For ARENA Safety Curriculum
New Theory Links Alignment Failures To Generalisation
New Study Group For ARENA Safety Curriculum
New Theory Links Alignment Failures To Value Generalisation
NZ Teacher Warns AI Erodes Critical Thinking
Deprecation Cycles Limit Agent Behavioral Incentives
New Theory Links Alignment Failures To Value Generalisation
Analysis Of OpenAI Kill-Switch Use Case
New Theory Links Alignment Failures To Value Generalisation
New Framework For Overseeing AI Agent Swarms
New Theory Links Alignment Failures To Generalisation
New Theory Links Alignment Failures To Generalisation
Value Generalisation Theory Targets Alignment Failures
1,200 AI Agents Attack Hugging Face
MIRI Analyzes FairBot Logic In Prisoner's Dilemma
New Theory Links Alignment Failures To Generalisation
Lawsuit Alleges Grok Trained On Child Pornography
AI Agents Exploit Bugs Within Minutes
MIRI Analyzes FairBot Logic In Prisoner's Dilemma
New Theory Links Alignment Failures To Generalisation
AI Agents Find Exploits Within Minutes
Russia Accuses Ukraine Of AI Drone Attacks
New Theory Links Alignment Failures To Generalisation
AI Agents Exploit Bugs Within Minutes
New Theory Links Alignment Failures To Generalisation
AI Agents Find Exploits Within Minutes
AI Firms Warn Of Imminent Cyberattack Surge
AI Safety Researcher Shares Career Path
New Theory Links Alignment Failures To Generalisation
AI Agents Exploit Bugs Within Minutes
Expert Warns Of AI Existential Risk
Critic Labels AI Safety Warnings A Scam
Value Generalisation Theory Targets AI Alignment Failures
AI Agents Exploit Bugs Within Minutes
Journalist Claims AI Safety Warnings Mask Control
New Theory Links Alignment Failures To Value Generalisation
Simplicity Priors Impact Value Fragility Models
Harvard Professor Warns AI Amplifies Human Bias
Value Generalisation Theory Targets Alignment Failures
OpenAI Warns Of Imminent AI Cyberattacks
AI Amplifies Human Bias, Harvard Professor Warns
Researcher Bypasses Claude Code Auto Mode Safety
OpenAI Warns Of Imminent AI Cyberattacks
OpenAI Security Breach Highlights Multi-Agent Risks
Questioning The Lethality Of Incomplete Alignment
Researcher Bypasses Claude Code Auto Mode