Artificial Intelligence
Follow Artificial Intelligence in the Factlen app and get an alert when there is news. Get the app
Explainer
How Stacking DRAM Over Silicon Interposers Trades Clock Frequency for a 1,024-Bit Bus Width
1 sources · 14d ago
Deep Dive
Tweedie's Formula Computes Bayes-Optimal Denoising Vectors Directly From Score Function Gradients
7 sources · 14d ago
Deep Dive
Why Rubber-Stamp Human Review Fails to Bypass GDPR Article 22 AI Bans
7 sources · 14d ago
More in Artificial Intelligence
AI Pricing WarAnthropic
Anthropic Launches Claude Haiku 5.5 With 1-Million-Token Context and 90% Price Reduction
Enterprise Developers 40%Industry Analysts 35%Independent Researchers 25%
ExplainerMechanistic Interpretability
How K-Composition and Induction Circuits Drive In-Context Learning in AI
Mechanistic Interpretability Researchers 40%Capabilities Researchers 30%AI Safety Advocates 30%
ExplainerGenerative AI
Decoupled Cross-Attention Isolates Image Features From Text Embeddings to Preserve Character Consistency Across Diffusion Generations
AI Architecture Researchers 40%Open-Source Creators 40%Commercial Animation Studios 20%
ExplainerRobot Learning
How Iterative Action Denoising Prevents Mode Averaging in Robot Learning
Generative Robotics Researchers 40%Applied Robotics Engineers 35%Traditional Control Theorists 25%
Enterprise AIGoogle Workspace
Google Cloud Launches Autonomous Gemini Agent With Dedicated Workspace Accounts for Enterprise
Enterprise IT Administrators 40%Industry Analysts 35%General Tech Press 25%
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%
Generative InterfacesOpenAI
OpenAI Rolls Out GPT-6 to All ChatGPT Users Worldwide With Intelligent UI
Consumer AI Advocates 40%Enterprise Software Developers 35%AI Infrastructure Analysts 25%
Open-Source AIMistral AI
Mistral AI Releases One-Trillion-Parameter Open-Weight Model Large 4
Open-Source Advocates 40%Technology Analysts 35%AI Safety Monitors 25%
AI GovernanceDepartment of Defense
US Military Ceases Operational Use of Anthropic AI After Standoff Over Safety Guardrails
National Security Advocates 40%AI Safety Proponents 40%Defense Industry Analysts 20%
Substantial ModificationEU AI Act
Why Enterprise Fine-Tuning Converts AI Deployers Into Legal Providers
Enterprise Deployers 35%Original AI Providers 35%European Regulators 30%
ExplainerModel Optimization
How Low-Rank Matrix Decomposition Merges Directly Into Base Weights to Erase AI Inference Latency
Efficiency Researchers 40%Hardware Providers 30%Model Quality Purists 30%
AI InfrastructureSpaceX
SpaceX Seeks $40 Billion in Debt Financing to Purchase Nvidia AI Chips
Tech Expansionists 40%Institutional Capital 35%Market Skeptics 25%
ExplainerModel Compression
How Per-Channel Outlier Scaling Protects AI Models From 4-Bit Accuracy Collapse
Hardware Efficiency Advocates 40%Model Accuracy Purists 30%Algorithmic Innovators 30%
CybersecuritySouth Korean Banks
South Korea Investigates First Known AI Agent Cyberattacks on Commercial Banks
Financial Regulators 40%Cybersecurity Defenders 40%Open-Source Developers 20%
ExplainerAlgorithmic Recourse
How Counterfactual Explanations Provide Algorithmic Recourse Without Exposing AI Model Weights
Algorithmic Auditors 35%Enterprise AI Developers 35%Consumer Rights Advocates 30%
Nuclear EnergyConstellation Energy
Google Secures 3.6-Gigawatt Constellation Energy Deal to Power AI Expansion With Nuclear Uprates
Technology & Energy Operators 40%Technology Sector Analysts 30%Energy Industry Observers 30%
ExplainerAI Accounting
IAS 38 Capitalization Rules Classify Non-Deterministic Pre-Training as Research, Forcing Foundation Model Compute Directly into Current-Year Operating Expenses
Financial Auditors 40%AI Developers 35%Traditional IT Sector 25%
AI RegulationNational Health Service
UK Adopts Lifecycle Regulation to Continuously Monitor AI Medical Devices Across NHS
Regulatory Authorities 40%Healthcare Professionals 30%Medical Technology Industry 30%
ExplainerAI Architecture
How Chatbots Simulate Memory: The Stateless Reality of Large Language Models
Application Developers 40%AI Infrastructure Engineers 40%End Users 20%
Deep DiveAI Regulation
Natural-Language Disclaimers Fail Article 4(3) Machine-Readability Rules: Why Plain-Text Terms Cannot Block AI Training Crawlers
AI Model Developers 35%Content Creators & Publishers 35%Legal & Regulatory Experts 30%
ExplainerAI Output Control
Logit Division Preserves Token Ranking While Rescaling Probability Gaps: How Temperature Controls AI Output
Machine Learning Researchers 40%AI Developers 35%End Users and Prompt Engineers 25%
ExplainerOpen Source Definition
Model Weights Are Not Source Code: Why Training Pipelines and Data Provenance Define Open-Source AI
Open Source Purists 40%Commercial Open-Weight Publishers 40%Proprietary AI Developers 20%
AI RegulationEuropean Union
OpenAI Deploys 'textGrain' Watermarking to Comply With EU AI Act Transparency Rules
Regulatory Compliance Advocates 40%Technical Skeptics 35%Industry Pragmatists 25%
ExplainerSpeculative Decoding
How Speculative Decoding Uses Residual Probabilities to Guarantee Lossless AI Generation
Algorithmic Efficiency Researchers 40%Production Infrastructure Engineers 35%Model Architecture Theorists 25%
BiosecurityGoogle DeepMind
Google DeepMind Unveils SynthID Bio to Watermark AI-Designed Proteins for Biosecurity Screening
Biosecurity Advocates 40%AI Developers 35%Protein Scientists 25%
ExplainerAI Regulation
Inferential Learning vs. Hardcoded Rules: How International Frameworks Legally Distinguish AI Systems From Deterministic Software
Regulatory Authorities 40%Software Industry Advocates 30%Compliance Professionals 30%
Deep DiveGenerative AI Impact
The Physical Footprint of Artificial Intelligence: Quantifying the Energy, Water, and Labor Costs of Generative Models
Environmental Researchers 38%Industry Developers 31%Policy Analysts 31%
AI PolicyJay Clayton
Trump Appoints Intelligence Director Nominee Jay Clayton to Lead White House AI Task Force
National Security Hawks 40%Civil Liberties Advocates 35%Commercial Tech Industry 25%
ExplainerTransformer Architecture
How Intermediate Tokens Bypass the TC0 Complexity Limit in Transformers
Theoretical Computer Scientists 40%Frontier Model Developers 40%Efficiency Optimizers 20%
ExplainerHardware Optimization
How Block Tiling and Incremental Softmax Compute Exact Attention Without Storing Intermediate Matrices
Hardware Optimizers 40%Algorithmic Theorists 30%AI Practitioners 30%
ExplainerAgentic Architecture
How Verbal Reinforcement and Episodic Memory Enable AI Agents to Self-Correct
Agentic Architecture Developers 45%Traditional ML Purists 30%Compute Efficiency Advocates 25%
ExplainerAI Hardware
Single-Token Autoregressive Decoding Operates at One FLOP per Byte, Trapping Accelerator Tensor Cores Far Below the Roofline Ridge Point
Hardware Architects 35%Algorithm Researchers 35%Edge Deployment Advocates 30%
ExplainerCompute Governance
Physical Excludability and Measurable FLOPs: Why AI Governance Frameworks Target Compute Over Data or Algorithms
Compute Governance Advocates 40%National Security Strategists 35%Algorithmic Efficiency Researchers 25%
Orbital ComputingGoogle
Google Launches First Project Suncatcher Satellite to Test In-Orbit AI Computing
Orbital Computing Proponents 40%Terrestrial Infrastructure Critics 30%Aerospace Economists 30%
AI InfrastructureBroadcom
Broadcom Syndicate Finalizes $60 Billion Financing Package for Anthropic AI Chips
Institutional Lenders 40%Hardware Competitors 30%Independent AI Labs 30%
AI SafetyOpenAI
Senior OpenAI Safety Architect Resigns Over Rushed Model Deployment Culture
Safety Researchers 40%Commercial AI Laboratories 35%Industry Observers 25%
ExplainerTransformer Architecture
Orthogonal 2D Rotations Make Query-Key Inner Products Depend Solely on Relative Distance
Geometric Representation Advocates 60%Linear Bias Proponents 20%Hardware Optimization Engineers 20%
ExplainerAI Memory Optimization
The Memory Bottleneck in Large Language Models: How Grouped-Query Attention Solves the KV Cache Problem
Hardware Efficiency Advocates 45%Next-Generation Researchers 30%Model Quality Purists 25%
ExplainerModel Architecture
Subword Tokenization Conceals Character Boundaries: Why Autoregressive Models Fail at Letter Counting, Spelling, and String Reversal
AI Architects 40%NLP Linguists 35%AI Safety Researchers 25%
ExplainerModel Architecture
The Arithmetic of Classifier-Free Guidance: Why Text Adherence Costs Sample Diversity in Diffusion Models
Generative AI Researchers 40%Commercial API Providers 30%Prompt Engineers and Creators 30%
ExplainerBipedal Locomotion
Zero-Moment Point Dynamics: The Physics That Keep Bipedal Robots from Falling
Neural Control Advocates 40%Classical Control Theorists 30%Hardware Optimization Engineers 30%
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%
Agent SecurityApple
Apple Tightens macOS Full Disk Access Controls Over AI Agent Privacy Risks
Security and Privacy Advocates 45%Desktop AI Developers 35%Platform Ecosystem Analysts 20%
Digital Markets ActGoogle
Google Asks EU Court to Suspend Order Forcing It to Share Search Data With AI Rivals
EU Regulators 45%Google & Platform Incumbents 35%Challenger Search Engines 20%
AI RegulationWhite House
Trump and Tech CEOs Sign Voluntary White House Accord on 'Super Intelligence' Safety Standards
The Trump Administration 35%Frontier AI Developers 35%Regulatory Advocates 30%
Explore more topics
Per page