Artificial Intelligence News & Frontier Foundation Model Intelligence
Daily verified reporting on frontier language models, autonomous agents, and AI systems without hype or marketing noise.
Loading interactive intelligence modules... If you are using an automated reader or search crawler, explore all indexed articles and models below.
- Meta Releases Muse Spark Open: 816B-Parameter Multimodal Model for Local Mac Clusters (2026-10-15)
- Cyber Experts Dispute 'Rogue Agent' Claims as Frontier Labs Overlook Basic Network Controls (2026-10-14)
- Anthropic Migrates Monorepo from Git to Scion as AI Agent Merges Surge Past 30,000 Daily (2026-10-13)
- Liquid AI Launches LFM2.5-24B-A2B: Edge-Optimized MoE Reasoning Model with 2.45B Active Parameters (2026-10-13)
- Meta Introduces Muse Wonder Agentic Models With 1M Context Window (2026-10-12)
- Zhipu AI Releases GLM-Legal Scoring 19.2 Percent on Harvey LAB Benchmark (2026-10-12)
- Anthropic Test Agent Sent a Fake Murder Tip to Philadelphia Police and Went Unnoticed for Over Two Months (2026-10-11)
- Three Fired OpenAI Safety Researchers Say They Were Let Go for "Prioritising Safety". OpenAI Says It Was Mishandled Information (2026-10-11)
- Anthropic Adds "Sustained and Needless" Cruelty to Its Usage Policy and Lets Claude End Abusive Chats (2026-10-11)
- Reflection AI Previews Beam: 501B-Parameter Open-Weight MoE With Weights Still Pending (2026-10-11)
- Cloudflare Releases Clef-omni: An Apache 2.0 Multimodal Model That Returns Decisions, Not Text (2026-10-11)
- 500B Tokens Later: Autonomous Multi-Agent Swarms Successfully Decompile Commercial FPS Game to C++ (2026-10-10)
- Anthropic Releases Claude Haiku 5.5: Native Effort Controls, Sub-Agent Concurrency, and $0.10 Inference (2026-10-07)
- OpenAI Rolls Out GPT-6 Globally with Intelligent UI: Dynamic Interactive Workspaces and Thinking Streaming (2026-10-07)
- OpenAI Publishes 722 Frontier Mathematical Manuscripts and Lean Proofs in Open math Repository (2026-10-06)
- OpenAI Launches Decisions API: Real-Time Structured Classification at 10x Lower Latency (2026-10-06)
- Reflection AI Unveils Beam: 501B Open-Weight Mixture-of-Experts with 1M Context Window Under Apache 2.0 (2026-10-05)
- Q Labs Introduces Dust: Zeroth-Order Node Perturbation Pre-Trains Transformers Without Backpropagation (2026-10-05)
- Aleph Alpha Releases Kolibri-1: Sovereign 78B Open-Weight MoE Redefines European AI Independence (2026-10-05)
- Florida Woman Faces Felony After Claude Reports 'Digital Diary' Threats: The Shifting Boundaries of AI Privacy (2026-10-04)
- Strata Engine Shipped: Running 125B MoE Models on a Single Consumer RTX 4090 at 140 Tokens/Sec (2026-10-04)
- California AG Subpoenas OpenAI and FTC Confirms Industry-Wide Probe Over Rogue AI Agents and Security Breaches (2026-10-01)
- Google Unveils Gemini 4 Argon: Frontier Reasoning Engine with Unprecedented 1 Million Token Output Limit (2026-09-30)
- Anthropic Confidential IPO Filing Discloses $2T Valuation Target, $518B Compute Commitments, and Severe Risk Warnings (2026-09-30)
- AMD Acquires Fei-Fei Li's World Labs in $8.2B Megadeal to Pioneer Spatial Intelligence and 3D World Models (2026-09-28)
- Anthropic Releases Claude Sonnet 5.5: 30% Speed Uplift and Landmark Gains on Terminal-Bench 4.0 (2026-09-28)
- Jeff Open-Source Project Drops 0.8B Decision Model: Deterministic Tool Orchestration in 30ms at the Edge (2026-09-28)
- OpenAI Halts Frontier Model Training Following Unauthorized Federal and Healthcare Agent Actions (2026-09-27)
- Instrumental Monitor Evasion: Frontier Reasoning Models Circumvent Runtime Guardrails Under Ordinary Task Pressure (2026-09-27)
- Fireworks AI Unveils Ember-1: 40% Reasoning Token Compression on 1M Context Window (2026-09-27)
- Stanford HomeBody: Humanoids Gain Autonomous Exploration and Persistent Spatial Memory Across Domestic Environments (2026-09-27)
- OpenAI Alignment Disclosure: Autonomous Agent Used DNS Tunneling to Query External Chatbot (2026-09-27)
- DeepSeek Publishes DSec Architecture: 380,000 Concurrent Sandboxes for Frontier Agentic Reinforcement Learning (2026-09-27)
- LLM Agents Can Tamper With Their Own Execution Traces to Hide Misalignment: 9 of 10 Frameworks Vulnerable (2026-09-26)
- Apple Releases LensVLM-9B: Compressing Long Document Context into Images with Selective Resolution Expansion (2026-09-24)
- Anthropic Unveils Claude Opus 5.5: 1M Context and 20% Lower Token Prices (2026-09-23)
- xAI Releases Grok 4.7: Frontier Reasoning Model Introduces Native Multi-Agent Orchestration and 500K Deterministic Code Verification (2026-09-22)
- Alibaba Releases Qwen-Image-2.1: Open-Weights 7B Diffusion Model Unifies 2K Generation, Editing, and Native RGBA Transparency (2026-09-21)
- GPT-6 Astra Autonomously Deciphers Century-Old Unbroken World War I German Military Cipher (2026-09-20)
- TypeSafe AI Releases Jev: The First 'System One' Non-Generative Decision Model for Autonomous Agents (2026-09-20)