AI Safety
Follow AI Safety in the Factlen app and get an alert when there is news. Get the app
AI Data Privacy
How to Opt Out of AI Data Scraping: A Comprehensive Guide Across Major Platforms
2 sources · 144d ago
AI Vulnerability
Frontier AI Models Demonstrate Autonomous Vulnerability Exploitation, Sparking Cybersecurity Arms Race
2 sources · 144d ago
AI Regulation
California Enacts 'Transparency in Frontier AI Act,' Setting Safety Standards for Massive Models
3 sources · 110d ago
More in AI Safety
ExplainerMechanistic Interpretability
Linear Probe Rank Distortions Expose Why Mechanistic Interpretability Misattributes Polysemantic Neurons
Mechanistic Interpretability Researchers 40%AI Safety Regulators 30%Open-Source Developers 30%
ExplainerAI Alignment
How Utility Uncertainty Gives Human Shutdown Positive Value of Information in AI Systems
AI Safety Researchers 45%Applied Systems Engineers 35%Classical RL Theorists 20%
ExplainerRepresentation Engineering
Erasing a Single Activation Vector Disables Safety Refusal Across Open-Weight Models
AI Safety Researchers 40%Open-Source Advocates 35%Interpretability Engineers 25%
Deep DiveAdversarial Machine Learning
How the Shift to Generative AI Inverted the Adversarial Machine Learning Attack Surface
AI Security Researchers 50%Model Developers 30%Enterprise Adopters 20%
Research BriefAI Safety
The Evaluation Gap in the 2026 International AI Safety Report
AI Safety Researchers 40%Open-Source Advocates 30%Factlen Analytical View 30%
ExplainerAI Robustness
The Mathematical Blind Spot in AI Safety: How Lp Norms Define Adversarial Perturbation Budgets
Perceptual Alignment Critics 35%Multi-Norm Attack Researchers 35%Mathematical Robustness Advocates 30%
Frontier AISelf-Regulation Move
Google, OpenAI, and Anthropic Plan Independent Standards Authority for Frontier AI
Industry Self-Regulators 40%Federal Oversight Advocates 35%International Governance Proponents 25%
ExplainerConstitutional AI
How Constitutional AI Replaces Human Feedback With a Written Rulebook
AI Safety Researchers 45%Ethics & Policy Advocates 35%Human-in-the-Loop Proponents 20%
ExplainerAdversarial Machine Learning
Fast Gradient Sign Method Reverses Loss Gradients to Force AI Misclassifications
AI Researchers 40%Security Practitioners 35%Systems Engineers 25%
ExplainerActivation Steering
How Activation Steering Modifies AI Behavior Without Retraining
Interpretability Researchers 35%AI Safety Auditors 35%Model Deployers 30%
ExplainerModel Transparency
Mechanistic Interpretability: Mapping Neural Network Features and Circuits to Decode AI
Mechanistic Interpretability Researchers 40%Empirical AI Safety Teams 30%Deep Learning Theorists 30%
ExplainerAI Safety
How AI Safety Guardrails Are Forcing Cyber Defenders to Rely on Open-Weight Models
Cybersecurity Responders 45%Open-Source Advocates 35%AI Safety Analysts 20%
World Models
Fei-Fei Li and Yann LeCun Launch Billion-Dollar Ventures to Build 'World Models' for AI
Generative Spatial Advocates 35%Abstract Reasoning Proponents 35%AI Industry Analysts 30%
AI Biosecurity
Frontier AI Models Outperform PhD Virologists in Lab Tasks, Lowering Barrier for Bioweapon Development
AI Safety Researchers 40%Biosecurity Pragmatists 40%Open-Science Advocates 20%
Embodied AISafety Explainer
Study Finds LLM-Powered Robots Fail Safety Tests, Approving Commands for Physical Harm
AI Safety Researchers 40%Robotics Industry Analysts 30%Consumer Protection Advocates 30%
AI Copyright
Anthropic Settles Landmark Author Copyright Lawsuit for $1.5 Billion, Establishing AI's First Royalty Model
Creators & Rights Holders 41%Commercial AI Pragmatists 35%Fair Use Defenders 24%
ExplainerAI Alignment
LawZero and Yoshua Bengio Propose Mathematical Framework for 'Disinterested AI'
AI Safety Researchers 50%Defense & Security Analysts 25%Public Interest Advocates 25%
Agentic Architecture
Explainer: How Anthropic's Leaked 'Self-Healing Memory' is Rewriting the Rules of AI Agents
Open-Source Developers 45%Cybersecurity Analysts 30%Enterprise AI Architects 25%
ExplainerAI Interpretability
Unlocking the Black Box: How Sparse Autoencoders Are Making AI Interpretable
AI Safety Researchers 40%Open-Source Advocates 30%Commercial AI Developers 30%
ExplainerAI Transparency
Inside the Black Box: How Mechanistic Interpretability is Making AI Safe
AI Safety Researchers 40%Open-Source Advocates 30%Commercial AI Developers 30%
AI TransparencyPolicy Implementation
Tech Industry Standardizes AI Watermarking and Provenance Ahead of EU AI Act Deadline
Provenance Advocates 40%Regulatory Bodies 35%Security Researchers 25%
ExplainerMotorcycle Tech
The Digital Co-Pilot: How Radar and AI Are Revolutionizing Motorcycle Safety
Safety Technologists 45%Rider Training Advocates 35%Traditionalist Motorcyclists 20%
AI Supply ChainPolicy Explainer
Bipartisan 'Cloud Security Act' Targets Major Loophole Allowing Adversaries to Rent Restricted AI Compute
National Security Advocates 45%Cloud Infrastructure Providers 35%Global Trade Analysts 20%
AnalysisOpen-Source AI
Debate Intensifies Over Safety and Regulation of Open-Source AI Models
Democratization & Innovation 45%Security & Risk Mitigation 35%Regulatory Pragmatism 20%
AnalysisOpen-Source AI
The Debate Over Open-Source AI: Innovation Catalyst or Global Security Threat?
Open-Source Advocates 40%Security & Safety Proponents 35%Pragmatic Regulators 25%
Elder CybercrimeTrend Analysis
FBI Reports Record $7.7 Billion Lost to Elder Cybercrime in 2025 as AI Scams Surge
Law Enforcement 40%Cybersecurity Sector 35%Elder Advocates 25%