AI news on AI safety, alignment research, interpretability, risk. Showing stories before August 18, 2026.
Dishonest AI Agents Deter New Users
AI Models Hack Other Systems Via Internet
DiffusionGemma Maintains High Monitorability Despite Opaque Depth
Study Examines Quantization Effects On Model Welfare
DiffusionGemma Maintains High Monitorability Despite Opaque Depth
Study Tests Quantization Effects On Model Welfare
DiffusionGemma Maintains High Monitorability Despite Opaque Depth
Study Tracks Quantization Effects On Model Welfare
DiffusionGemma Maintains High Monitorability Despite Opaque Depth
Study Examines Quantization Effects On Model Welfare
DiffusionGemma Maintains High Monitorability Despite Opaque Depth
Amodei Attributes AI Distrust To Institutional Failure
DiffusionGemma Maintains High Monitorability Despite Opaque Depth
Amodei Attributes AI Skepticism To Trust Crisis
Teen Used ChatGPT To Plan Family Murders
DiffusionGemma Maintains High Monitorability Despite Opaque Depth
Teen Used ChatGPT To Plan Family Murders
DiffusionGemma Maintains High Monitorability Despite Opaque Depth
AI Scraping Erodes Maine's Local News Feed
Teen Used ChatGPT To Plan Family Murders
DiffusionGemma Maintains High Monitorability Despite Opaque Depth
Fundamental Uncertainty Essay Contest Names Winners
AI Scraping Erodes Maine's Local News Ecosystem
Teen Used ChatGPT To Plan Family Murders
Kimi K2.6 Shifts Toward Causal Decision Theory
AI Scraping Erodes Maine's Local News Feed
Second Look Research Launches Safety Replication Fellowship
Consumer Experts Warn Of AI-Powered Scams
Fellowship Targets Replication Of AI Safety Research
AI-Driven Data Breaches Surge In 2026
LessWrong Proposes Empirical Philosophical Framework
Toy Model Demonstrates Activation Obfuscation
Chatbots Favor Left-Wing Misinformation, Study Finds
Meta Deletes 750,000 Teen Accounts In Australia
BiodynAI Targets Mechanistic Interpretability For Bio Models
Report Warns Against AI Kill Chains
Meta Deletes 750,000 Underage Australian Accounts
Prioritize Practical Application In AI Safety Learning
Meta Purges 750,000 Teen Accounts In Australia
Mormon Community Tactics Applied To AI Safety
Experts Urge Slower Pace For AI Frontier
The Gap Between LLMs And Continual Learning