AI Agents & Automation

Preliminary report links Amap agent queries to Tencent Cloud infrastructure

Researchers report parallel AI agent queries for Amap entrance-navigation data and traces tied to Tencent Cloud infrastructure. The operator and purpose remain unknown.

How to try DeepSeek Harness in a disposable workspace

A documentation-based first-use guide to DeepSeek Harness in a disposable workspace, covering model setup, provider charges, session data and preview safety limits.

Google pauses new open-source product vulnerability reports: where researchers can still report

Google paused new OSS VRP product vulnerability submissions on October 1. Supply chain reports and existing cases continue, while some findings may fit separate Cloud, AI or Patch Rewards rules.

SageMaker multi-turn RL: what a search team must verify before training

AWS reports gains on three of four held-out search benchmarks for its Qwen3.6-27B agent. This guide maps the SageMaker setup, costs and evaluation needed before a trial.

Runway’s Praxis-1 tests a video-trained route to robot control

Runway says Praxis-1 applies video pretraining to robot control. Its placement-error comparison offers a limited signal while public weights and fuller evaluation details remain pending.

How AWS's lease review sample separates AI from compliance decisions

AWS's sample routes lease questions through Amazon Quick to fixed tools and a rules engine. The guide explains the receipt, evidence trail, setup limits and what adopters must validate.

Claude Code mods: create a context display and inspect its access

Claude Code mods can draw inside a session and change its behavior. This documentation-based guide explains how to request a context display and inspect its access.

DoorDash’s AI launches span text ordering, Dasher help and restaurant tools

DoorDash's September 30 AI launches cover text ordering, Dasher support, restaurants and office agents. Access ranges from live SMS help to selected-user and corporate betas.

OpenAI and Synopsys plan a chip-design model that can use EDA tools

The companies agreed to develop GPT-Synopsys around chip-design tools and report early customer engagements. The announcement gives no public launch date, price or measured result for the model.

Meta brings Muse into small-business accounts and tools

Meta has added business accounts and outside tools to Muse. Owners still need to approve publishing, sending and spending, while access and plan limits vary.

Set up an Agents API browser task with OpenAI computer use

A documentation-based guide to creating an OpenAI Agents API browser session, handling website-origin requests, checking the result and deleting the session.

Continues OpenAI DevDay 2026: launches, availability and what builders can use

How to set up a shared Team Task in ChatGPT

OpenAI’s Team Tasks let eligible ChatGPT Business and Enterprise teams manage recurring work from a shared workspace. This guide explains setup, account permissions, run review, pause controls, and what remains plan-specific or undocumented.

Set up a Codex Cloud environment for a repository

Connect a GitHub repository, publish a reusable Codex Cloud environment and learn how isolated tasks continue across desktop, web and mobile. This documented guide explains access, network, credentials and cost; BIG CHANGE has not tested it.

Manus 2.0 adds video editing, game tools and personal agents

Manus 2.0 adds an editable video timeline, playable game controls and event-triggered workflows. See access, credits and persistent hosting costs.

OpenAI holds GPT-6.1 Astra after scope and authorization concerns

OpenAI has delayed GPT-6.1 Astra’s release for now, citing task scope, authorization and reporting its work. Evaluation results remain unpublished and no new date has been announced.

Shopify adds browser agent tools inside checkout, with buyer approval required

Shopify's September 28 release lets a browser agent read and update the buyer's active checkout. Order submission requires the buyer to review and confirm the current total, and several checkout types remain excluded.

NVIDIA's agent safety platform has a software boundary and an optional hardware watchdog

OpenShell puts access controls around AI agents. NVIDIA's optional Sentry design adds monitoring on BlueField-4, though its speed and effectiveness remain vendor claims.

Multiverse says its Compass AI reviews every instructor session

Multiverse says its Compass AI reviews instructors' sessions for quality and suggests development areas. Teachers told The Guardian that the system feels like constant scrutiny; the company says human managers make employment decisions.