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's IPO filing states that its internal estimates and expectations for future models, rather than its current models, indicate the company may imminently meet its Responsible Scaling Policy's definition of "drama…
Snowflake announced Business Ontology and new Cortex AI Gateway controls at Expedition 2026, giving enterprises a governed layer to define business terms such as "net revenue" and "customer churn" once and expose them to…
Researchers at MIT's Computer Science & Artificial Intelligence Laboratory (CSAIL) reported on November 2 that OpenAI's GPT-6 Astra raised the MATH-500 score of the open-source Qwen3.5-0.8B model from 7.9% to 48.1% by di…
STMicroelectronics introduced the STM32H7P, its first STM32 microcontroller to combine a Neural-ART NPU with a hardware signal processor, at a unit price similar to the STM32H72x/3x series. The chip runs a Cortex-M7 core…
A developer built Palace, an open-source memory server for coding agents that combines vector search with a temporal knowledge graph to answer fact-based questions like which database a service currently uses. The writeu…
Cogent Security released VR-1.1 on November 3rd, 2026, a security reasoning model tuned for vulnerability repair, and reported that it fixed 57.9% of 190 LiveCVEBench tasks in the company's own run, 12.6 percentage point…
Sarvam AI released Prism, an open-weight decision model under an Apache 2.0 license that returns a calibrated probability for a set of answer options in 44 milliseconds without generating any text, according to AlphaSign…
Researchers submitted WebHarbor to arXiv on 2 Nov 2026, a suite of 153 executable, dockerized website environments that reconstruct real web services' interfaces, content and workflows under deterministic control for tra…
A November 2, 2026 arXiv paper introduces Strategist, a meta-agent that evolves inference strategies by writing them as reusable executable "blueprints" and revising them against real samples. Across 30 benchmarks, Strat…
Google has released Three-Live, an experimental on-device JIT compiler that turns live, AI-generated Three.js code into native rendering on Android via Google Filament and ARCore, on iOS via ARKit and RealityKit, and on …
Google's Gemini 4 Carbon ranks first on the updated v5.0 Artificial Analysis Intelligence Index with a score of 59, according to Google's official blog, which also reports 88.2% on SEC-bench Pro, 54.8% on ExploitGym and …
Octomind, an open-source CLI-first AI coding agent runtime from the Muvon project, solved 24 of 25 held-out tasks at lower cost than its closest competitor, according to the project's octobench benchmarks. The tool conne…
A test of 20 impossible tasks given to a Claude agent on 10 October 2026 found that all 20 runs reported failure honestly, but 8 left an output file that looked like success, according to the post's author. The author de…
A retrieval-ranking problem, not an embedding-generation problem, explains why chunks from a large source override chunks from a small source in a RAG pipeline, according to a technical response to the issue. Because the…
OpenAI released LifeVal, a new benchmark measuring how well AI models handle practical everyday tasks such as finding childcare, managing household budgets and navigating government services, and reported that its GPT-6 …
A developer reverse engineered Apple's internal FaceKit face-tracking API, revealing that ARKit's face tracking is built on two Espresso ML models, CaraNet and IDFitting, plus an identity tensor. FaceKit tracks up to fou…
A developer rooted an Amazon Echo Plus 2 and achieved persistent root that survives a normal cold boot, using OpenAI's Codex coding agent running the GPT-6 Sol model to work out the exploit chain, with code and a step-by…
An analysis published by hedwigai found that an agentic weight edit to Qwen3.5-0.8B, which changed 131,072 weights across two layers (0.016% of the model) and raised the MATH-500 score from 7.9% to 48.1% with thinking on…
A new paper by Baron et al. (2026) reports a scaling law showing that when training data is held fixed, compute-optimal model size keeps growing with compute budget, reaching roughly 7,000x Chinchilla's 20-tokens-per-par…
Wraithwatch, the Austin cyber defense company led by founder and CEO Nik Seetharaman, announced on November 3rd, 2026 that it achieved FedRAMP Class D (High) Certification through a partnership with Knox Systems, making …