AI news for builders and product teamsUpdated Nov 18, 2026, 07:00 UTC
AI news, straight from the source.
Releases, research, and ideas for developers and product teams. Updated every hour.
Across the feeds
Newest first
An AWS Machine Learning Blog post details how to build ambient agents using Amazon Bedrock AgentCore, where event streams like Amazon S3 uploads trigger agent jobs that pause for human input via a single ask_human tool. The reference sample combines AgentCore Runtime, Lambda, DynamoDB, SQS, and a React frontend into a serverless platform.

An AWS Machine Learning Blog post details a step-by-step guide for giving three environment types access to Claude Platform on AWS inference: AWS workload accounts via cross-account SigV4, developer laptops via a workspace-scoped API key, and external workloads via OIDC federation, all sharing one subscription with workspace isolation.

Amazon Quick Sight announced the hierarchy filter, a single control that lets dashboard authors present multi-level drill-down filtering instead of separate controls. It supports up to five levels, dimension fields only, and searches limited to the highest hierarchy level.

Researchers from Carnegie Mellon, MIT, NYU, and Stanford built an AI called Ataraxos that beat top Stratego player Pim Niemeijer 15 games to one, with four draws, in 20 online games. It trained on 16 GPUs for a week and a few thousand dollars.

Albertsons Companies is using ChatGPT Enterprise and the OpenAI API to help its teams work faster and to make grocery shopping easier for millions of customers, according to OpenAI.

Cloudflare released two decision models it trained itself, Clef and Clef-flash, on Workers AI, open-sourced them on Hugging Face under Apache 2.0, and launched an early reinforcement-learning fine-tuning service.

uniopen customized Amazon Nova 2 Lite for its retail content moderation policies using supervised fine-tuning with LoRA in Amazon SageMaker AI plus a prompt-level output change. Per Behavior Macro F1 rose from 0.5852 to 0.8550 and Subject Type Macro F1 from 0.4162 to 0.8491, both above production targets.

Airbnb CEO Brian Chesky said in a TechCrunch interview that chatbots are not the right interface for e-commerce and travel, and argued the industry lacks a true AI-native operating system. Airbnb plans to explore "multiplayer" AI interfaces over the next three to six months and roll out voice agents this fall.

Hugging Face released Olmo-core 3, an open training framework for large mixture-of-experts (MoE) models, designed to scale MoE training into the trillion-parameter range while preserving computational efficiency. It is part of the systems behind the next generation of Olmo.

Google says Gemini 4 Argon is a frontier model with competitive benchmarks and introductory pricing of $2 per million input and $10 per million output tokens, but it is not yet available to use. OpenAI meanwhile shipped GPT-6.1 Sol at the same price after pulling a stronger Astra candidate over alignment failures.

At OpenAI's annual DevDay conference, CEO Sam Altman announced Dots, a personal assistant agent powered by GPT-6 Astra. The Verge reports the launch is aimed at competing with Meta's Muse AI agent platform, with a demo showing Dots building metaverse-esque virtual worlds.

Anthropic is offering Claude for Government to US federal and state agencies in a FedRAMP High environment, after an open beta since July. The Pentagon will not use it because it still classifies Anthropic as a supply chain risk.

A Verge investigation details how Kevin O'Leary's proposed 40,000-acre, nine-gigawatt Utah data center project, called Stratos or Wonder Valley, was halted by local backlash and litigation. The project ran through the Military Installation Development Authority before county officials knew about it.

Photon, a startup that helps developers build AI agents for messaging platforms like iMessage and WhatsApp, has raised $4.5 million in seed funding. The company says it has over 40,000 developer sign-ups and grew revenue 10x in four months.

At OpenAI DevDay 2026, OpenAI launched Dots, always-on personal agents built on a never-ending GPT-6 Astra chat, alongside GPT-6.1 Sol, a Pro 500 plan, Sign in with ChatGPT, Plugins, ChatGPT Space and an Ultrafast mode. Google released benchmark scores for Gemini 4 Argon, which is not yet available to use.

Glow Security found more than 13,000 internal company screenshots from 343 organizations posted to public GitHub repositories by AI agents. The images reportedly showed customer data, login credentials, and unreleased product details.

Cloudflare is bringing two publicly built European open models, EuroLLM and Apertus, to Workers AI with access requests open now, and launching hands-on AI security workshops for government cyber agencies, starting in Singapore in October.

NVIDIA announced that 25 new games will join GeForce NOW throughout October, with six available the week of the announcement. The Witcher 3: Wild Hunt – Remastered launched Sept. 29 and is streaming for Performance and Ultimate members, who can also claim a CONTROL Resonant reward from Oct. 1 to Nov. 1.

NVIDIA published a blog post arguing its AI factories are engineered to maximize return on investment through three factors: earning capacity, useful life and demand, which it labels productive, durable and fungible. The post cites SemiAnalysis AgentX data and third-party analyses of GPU resale and rental values.

Legato, a hearing tech startup founded in 2024, has launched its AI-based hearing glasses, called Legato Frames, starting at $999 for adults with up to moderate hearing loss. The glasses, which process audio on-device with no cameras or recording, are available in the U.S. and eligible for vision insurance coverage.