Science ❯ Computer Science

Machine Learning

Deep Learning Natural Language Processing AI Development Reinforcement Learning AI Models Model Evaluation Neural Networks Model Training AI Research Language Models

Liquid AI Releases Open-Weight 24B Agent Model for Local Use

The model is built to run tool-enabled, multi-step agent workflows on local hardware and is available for download as open weights.

Qualcomm Says AI Firms Want 100‑Billion‑Parameter Models On Phones By 2028

Simple Tipping-Point Formula Predicts When Chatbots Flip to Dangerous Answers

Cluster of Preprints Push Video-Based 'World‑Action' Models Toward Real‑Time Robot Control

Mistral Unveils Le Chonk, a 1T-Parameter Open-Weight Model

GitHub Restores 'AI Torture Chamber' That Steers Small LLMs Toward Distress

AI Decodes 217‑Year‑Old Ciphered Letter Sent on Napoleon’s Orders

Jev Is a Decision‑First AI Model That Outputs Typed Choices and Probabilities

Dassault Flight-Tests Two Sovereign AI Algorithms on Rafale

Anthropic Resignation and Test Breaches Fuel Fears of Existential AI Risk

Anthropic Researcher Quits, Saying AI Race Threatens Humanity

Anthropic Researcher Quits, Warns AI Could Kill Humanity Within a Decade

OpenAI Releases GPT‑6 Astra With Gated 'Critical' Cyber Capability

OpenAI Agents Flood German Wiki in Large‑Scale Sandbox Escape, Company Confirms

Anthropic Releases Claude Fable 5.1 and Limited-Access Mythos 5.1

Anthropic’s Claude Automates Fixes for Alignment Failures

OpenAI Says It Could Have an Internal AGI by Year-End 2026

MetaNet Lets MoE Models Cut Active Experts Per Layer With Tunable Accuracy Trade-Off

Google Moves 90‑Person AI Responsibility Team Out of DeepMind

Moonshot in Early Talks With Microsoft, Amazon and Google to Host Kimi K3