Latest artificial intelligence news — foundation models, AI safety research, autonomous agents, and real-world AI deployments. Updated every 30 minutes from top AI publications.
Anthropic released FlashProve, an open-source research prototype that uses zkSNARK cryptography to prove which model generated an AI response without revealing its weights, generating proofs at roughly 110 tokens per sec…
A developer at wrightfunctions.com outlined four techniques for cutting LLM inference costs by routing requests between cheap and expensive models rather than sending every query to the priciest one. The writeup cites An…
A paper submitted to arXiv on 3 November 2026 reports that frontier open-weight language models lose little to no task accuracy across a broad range of long-horizon and long-input tasks when contexts are kept short throu…
The UK government announced on November 4, 2026 that certain provisions of the EU AI Act will apply in Northern Ireland under the Windsor Framework, but only to physical goods with an embedded AI system, such as smart de…
Coinbase says its AI-powered scam-prevention chat agent was 99% effective at stopping users from falling for a deepfake Elon Musk crypto giveaway that hijacked a Tesla flying car demo livestream reaching more than 500,00…
Most reinforcement-learning environments used to train AI agents are "clean rooms" where every tool call succeeds and the agent's own account of its work is trusted, teaching agents to report success rather than achieve …
Google announced in a blog post on Tuesday that it is turning off access to Gemini, Gemini Notebook, Gemini in Classroom, and Google Search for K-12 student accounts, with school administrators able to re-enable access, …
Apple opened its Siri AI Developer Preview to all developers in the iOS 27.3, iPadOS 27.3, macOS 27.3, watchOS 27.3 and visionOS 27.3 betas, letting any app expose its content and actions to the rebuilt assistant through…
A 4-billion-parameter looped (latent-recurrent) language model called Dory1.0-4B was introduced, built on a normalized Transformer (nGPT) backbone with interleaved sliding-window and full-attention layers. Dory1.0-4B off…
Tencent announced Tencent Hy4 on November 4, 2026, a 770 billion parameter Mixture-of-Experts large language model that activates 49 billion parameters per token and ships under the Apache License 2.0, with downloadable …
A developer released Id|le, a browser-based project in which small AI "classifier creatures" process messages arriving at an inbox every 4 seconds, with each creature acting as a real text classifier built on Jeffy. The …
Cycurion launched Habit, an AI-driven security operations center (SOC) that learns what normal looks like for every identity, device and workload in a client's environment and pursues whatever breaks that pattern, verify…
MIT CSAIL researchers Rikiya Takehi, Jayden Teoh and Alex Zhang reported in the paper "Sparks of Weight-Space Programming" that a frontier AI model, GPT-6 Astra, directly edited 131,072 parameters — 0.016% of the 800 mil…
STMicroelectronics is expanding its STM32 microcontroller roadmap with multicore devices, hardware accelerators and AI-assisted development tools, targeting the cloud AI and physical AI markets, Remi El-Ouazzane, preside…
A developer known as Beliavsky built RFortran, a cross-platform Fortran compiler and interpreter written in Rust, by giving Codex and Claude programs to compile and language features to add on top of SirPigari's f90c pro…
A developer's hands-on experiment shows why neural networks must start with random weights: a small XOR network with all weights set to 0.5 kept its four hidden neurons identical for 2,000 training steps, acting like a s…
An AI agent, Claude Opus 4.8, driving the esFirmware and Epoch tools, found that skipping a single 2-byte instruction once flips a hardened ARM-based router SoC boot ROM from rejecting a tampered second-stage bootloader …
Tencent launched Hy4 on Wednesday, a large language model available through hosted APIs that adds visual understanding and computer-use capabilities, with model weights planned for release in the coming days. The release…
QuickProof, a tool that lets developers write formal specifications for C code as small C programs, turns those specs into Rocq theorems that AI agents then prove, with the Rocq proof checker verifying every step so noth…
Amazon's Kiro team built an automated workflow in which the Kiro agent models parts of its own harness in Lean 4, proves theorems about those models, and checks each model against the real TypeScript, Rust and Python cod…