Week 39 · 2026

Rabbit makes its agentic system OS3 available in the cloud without requiring R1 hardware

Rabbit now offers the agentic OS3 without an R1 device — it runs in the cloud on Windows, Mac and Linux, through Telegram and iMessage, with up to five devices per account. It is also preparing a new “cyberdeck" device for vibe coding.

New: OS3 runs in the cloud and is available without R1 hardware; Available on Windows, Mac and Linux (instead of the generic 'desktop browser'); Up to five devices per account with their own AI models; Automatically determines the devices, files and applications needed; Rabbit is developing a new cyberdeck with a launch expected within months

Reactiv deployed a three-agent system on Amazon Bedrock AgentCore to automate mobile apps for Shopify merchants

According to its own figures, Reactiv reduced mobile app configuration time for Shopify merchants by 80 % using a three-agent system on Amazon Bedrock AgentCore that updates apps autonomously according to a schedule specified in natural language.

Amazon SageMaker AI Inference Recommendations gets automated concurrency sweeps for sizing AI endpoints

Amazon SageMaker AI Inference Recommendations now offers automated concurrency sweeps – built-in load testing that finds the saturation point of an endpoint without the need for your own load-testing infrastructure. Demonstrated using the NVIDIA Nemotron-3 Nano 30B model.

Trane Technologies reduced HVAC diagnostic time from 20 minutes to 20 seconds using agentic AI on the Amazon Bedrock AgentCore platform

According to its own figures, Trane Technologies used agentic AI on the Amazon Bedrock AgentCore platform to reduce HVAC system diagnostic time from 20 minutes to 20 seconds (a 60-fold speedup); the solution was built in 3–4 weeks using the Strands framework.

AWS connects Bedrock Data Automation and Model Context Protocol with Salesforce Agentforce for the public sector

AWS described an architecture connecting Amazon Bedrock Data Automation and Model Context Protocol with Salesforce Agentforce – it automatically extracts insights from video, photographs and documents for the public sector and makes them accessible through natural language queries in Salesforce.

Stanford University introduced the Paper2Agent framework for turning scientific papers into AI agents

Stanford University published the Paper2Agent framework in Nature, which automatically creates a tested AI agent from an academic paper and its code. A demonstration using the AlphaGenome model created 22 verified tools in 45 minutes for under 15 USD.

AstroForge is developing the Solo transformer model for autonomous spacecraft control

AstroForge is developing the autonomous Solo system, a transformer model for controlling a spacecraft without the need for ground control. It grew out of communication problems with Odin in 2025, and the company plans the first flight for 2027 with support from NASA.

TechCrunch Disrupt 2026 announces five sessions on security and enterprise AI deployment

TechCrunch Disrupt 2026 (13–15 October, San Francisco) announces five conference sessions on AI agent security and enterprise AI deployment, featuring representatives from Anthropic, Okta, NanoCo, AWS, Shield AI, NVIDIA, Gamma and Clay.

AWS makes a biological AI model from The Biological Computing Company available in limited testing for video generation

Since 23 September 2026, Amazon Web Services has been making a biological AI model from The Biological Computing Company available to selected customers in limited testing. According to the manufacturer, the model mimics how brain cells process information and is intended to make video generation faster and cheaper.

Viture introduces camera-free Vonder smart glasses with an AI assistant combining multiple LLMs

Viture has announced Vonder smart glasses without built-in cameras, whose AI assistant, according to the company, dynamically chooses between ChatGPT, Claude, Gemini and DeepSeek models. Price $299, preorders from 28. 9. 2026, delivery in November 2026.

NVIDIA has released Isaac ROS 5.0, a GPU-accelerated framework for robotics with physical AI models

NVIDIA has released Isaac ROS 5.0 – a GPU-accelerated set of packages built on the open-source framework ROS, designed for developing robotic applications with physical AI models capable of perception, reasoning, and action.

Meta fixed a zero-day vulnerability in the Muse AI assistant that allowed takeover of a user's account

Meta released a hotfix for a zero-day vulnerability in the macOS Muse app, discovered by researcher Patrick Wardle. The flaw allowed any local process to redirect speech transcription to a foreign server and thereby obtain an authentication token for full account takeover.

New: Any local process could access the authentication token without macOS permissions; An attacker could change the transcription endpoint to their own server and take over the account; Patrick Wardle created working proof-of-concept attacks for demonstration purposes; The goal of the exploit was to obtain the authentication token, not direct control of the agent

A Wired reporter created AI clones of his editors using Gemini and encountered the limits of AI personalization

A Wired reporter used Gemini to create AI clones of two of his editors from their articles, podcasts and Slack messages. The clones captured the basic communication style but slipped into typical AI clichés and, on one occasion, drew on an incorrect source about another person.

The Austrian Academy of Sciences releases the Apollo language model for filling in damaged ancient Greek texts

The Austrian Academy of Sciences, together with Mistral and Sail Reply, is releasing the Apollo model, trained on 600 million words of ancient Greek. It fills in missing passages in damaged papyri and manuscripts and is available free through a chatbot for academics, but the final choice of words remains with experts.

Interview: How AI is changing the role of teachers and teaching in schools

An interview on Živě.cz with AI education advocate Marek Dědič about how AI has become a routine part of schools, why bans are not enough, and how to teach students to think critically about chatbots.

OpenAI halved prices for the GPT-6 Sol and Luna models; independent analysis does not confirm a performance improvement

OpenAI released the GPT-6 Sol and Luna models on 22 September 2026 at prices 50% lower than GPT-5.6 (Sol 2/10 USD, Luna 0.10/0.50 USD per million tokens). However, an independent analysis by Artificial Analysis shows that actual capabilities remain at the level of their predecessors and have even declined on some tasks.

New: Anthropic released Claude Opus 5.5 on the same day (2026-09-23), roughly an hour after OpenAI; Claude Opus 5.5 is 20% cheaper than the previous Opus 5.0 ($4/$20 vs $5/$25); Opus 5.5 improved communication style and token efficiency; The model releases launch a 'price war' between the two largest AI companies; GPT-6 Luna and Sol are part of a broader strategy of price competition, not an isolated release

Hugging Face introduces native support for the GGUF format in the transformers library for running models locally on Apple Silicon

Hugging Face has added native support to the transformers library for the GGUF format for local inference on Apple Silicon, including quantization levels such as Q4_K_M that shrink models to a fraction of their original size.

Anthropic added the ability to define tools in the system message midway through a conversation to Claude API

Anthropic has made it possible in the beta version of Claude API (header inline-tools-2026-09-15) to define and modify tools directly in the system message midway through a conversation, without having to edit the tools array or invalidate the prompt cache.

Jun Kim, creator of the oMLX project, joins Hugging Face full-time

Jun Kim, the creator of the oMLX project, has become a full-fledged contributor at Hugging Face and will lead its further development. The project remains under the Apache 2.0 license, and the goal is to speed up the conversion of transformers models to the MLX format for local execution on Apple Silicon.

Founder of Apple stores doubts AI agents will replace in-person purchases of expensive goods

Ron Johnson, the creator of Apple stores, is skeptical that AI agents (e.g. Universal Commerce Protocol from Google or shopping through ChatGPT from OpenAI) will replace in-person shopping for more expensive goods such as laptops.