Daily AI news, sourcedUpdates / All
Updates Official · 968 items · newest issue first · page 19 of 49
AreaAllModelsAgents & codingApps & productsChips & infraResearchPricing & limitsBusiness & fundingPolicy & safetyOther
- NVIDIA researchers' AlphaNIST trains one robot assembly policy from binary rewards alone to 98.2% success on all 20 NIST board tasks in simulationOfficialarXiv (NVIDIA)Oct 15Research
- Tamnoon details Tami, its agent on Amazon Bedrock that investigates and remediates cloud security findings
- AWS publishes a sample Bedrock AgentCore agent system that builds PowerPoint decks from enterprise documents in under 90 secondsOfficialAWS Machine Learning BlogOct 15Agents & coding
- OpenAI says Third Moment Research runs 10,000 agents at once on 36- to 48-hour research sessions with Codex and GPT-6.1 SolOfficialOpenAI NewsOct 15Agents & coding
- LangGraph CLI 0.4.35 adds --image-pull-secret for deploying from a private container registry
- ComfyUI v0.40.0 adds partner nodes for Nano Banana 2.1, Vidu Q4 and Kling Virtual Try-On, and speeds up Qwen 3.5 on low VRAMOfficialGitHub @Comfy-OrgOct 14Apps & products
- llama.cpp adds --spec-draft-n-ubatch to set the draft context's batch size and save VRAM in speculative decoding
- Manus launches Team and Enterprise plans for businesses, with the Team plan starting at $20 per paid seat per month
- Hugging Face's tensor viewer now loads much faster on large sharded models
- Researchers hide an encrypted capability inside model weights that fine-tuning and other elicitation methods fail to unlockOfficialarXivOct 14Policy & safety
- OpenAI says Codex and ChatGPT Work now produce more than half of output tokens for its EU enterprise customers
- Anthropic opens a Singapore office, its fifth in Asia-Pacific, and signs an AI safety MOU with Singapore's digital ministryOfficialAnthropicOct 14Business & funding
- APUS AI open-sources Reflex Agent, a personal agent run by its own Reflex decision model, under the MIT license
- KoboldCpp 1.123 adds Qwen Image 2.1 support, approval modes for its built-in agent and full batched-mode featuresOfficialGitHub @LostRuinsOct 14Chips & infra
- NVIDIA publishes its Run:ai GPU orchestration platform for Kubernetes on GitHub under Apache 2.0OfficialNVIDIA Technical BlogOct 14Chips & infra
- NVIDIA says European cities are deploying agents and digital twins built on Cosmos, Nemotron and Omniverse to run city operations
- NVIDIA commits $250 million in compute over five years to UK agentic AI safety research and joins the Earthshot Prize
- Microsoft tests five reflection models for Foundry's agent optimizer and finds no single best choiceOfficialMicrosoft Foundry BlogOct 14Agents & coding
- UniMate generates text-conditioned motion for skeletons of any topology (SIGGRAPH Asia 2026)
- Kiro Web adds fallback models, per-step models in Workflows and conversation previews while sessions reopen