OpenAI's 'Silent Consensus' asks who keeps control as AI grows
OpenAI's Dean Ball published 'The Silent Consensus' on November 18, asking how AI could reshape society and who keeps control of institutions.
The top 0.01% of AI knowledge, and what it changes for your work
Daily AI and technology coverage — models, chips, agents and research, and what each development actually means for you and your business.
732 stories
OpenAI's Dean Ball published 'The Silent Consensus' on November 18, asking how AI could reshape society and who keeps control of institutions.
Hugging Face's harness adapters train a LoRA inside the agent's own runtime, lifting Qwen3.8-27B from 39% to 55% on Terminal-Bench while halving token
Google has folded Duplex phone-calling tech into Tasks in Gemini, letting the AI ring businesses and handle real-world chores on your behalf.
Google's new Gemini import feature lets users transfer chat history and preferences from other AI apps via a manual ZIP upload process.
Google Quantum AI researchers proved quantum contextuality survives at scale by winning a game classical computers cannot, across 105 qubits.
An arXiv preprint finds vision-language models selectively access concept information as tasks require, echoing global workspace theory from
OpenAI's new reference architecture guides security teams through deploying Daybreak Red models for supervised cyber investigations.
The open-source Bend 2 language, which uses compiler-enforced mathematical proofs to catch rule violations, is trending on GitHub with 25,635 stars.
torchtitan v0.4.0 aligns with PyTorch 2.15 and unifies pretraining, fine-tuning and RL in one codebase for teams training large AI models.
Ten frontier AI models on three shopping benchmarks showed no accuracy gain from extra reasoning, but their confidence rose significantly.
pnpm, a Node.js package manager storing dependencies once via content-addressable storage, hit GitHub's daily trending list on November 18, 2026.
OpenAI's new research finds coding agents speed up AI work but perform unevenly, and asks what that signals about recursive self-improvement.