Browse Artificial Intelligence News (383)

Kristen Womack reviews October’s Azure Developer CLI releases, covering deployment previews, Container Apps express support, environment value removal, and extension-aware pipelines. The recap also explains stable extension APIs, breaking Go API changes, reliability fixes, and new templates for deploying semantic search and streaming AI applications to Azure.

Waldek Mastykarz examines whether naming a workspace folder ‘eval’ changes GitHub Copilot CLI behavior. Across about 900 runs with four models, the name increased agents’ suspicion that they were being tested without producing statistically significant behavioral differences. The analysis explains the experiment’s controls, limitations, and practical guidance for evaluation harnesses.

amitch and Shehab Amin explain how Sail connects Spark-compatible workloads to Microsoft OneLake without copying data. They demonstrate lakehouse transformations, Python-based embedding generation, and scoped access through LakeSail Cloud Sessions, while outlining compatibility checks and limitations of the reported performance and cost comparisons.

Microsoft Threat Intelligence examines how North Korean threat actors use AI to accelerate reconnaissance, identity deception, malware development, and operational scaling. The report distinguishes observed activity from suspected AI use, explains why operations remain human-directed, and provides Microsoft Defender detections and practical guidance for protecting identities, endpoints, and software repositories.

nareshvenkat introduces Project Osmos, a data engineering agent for Microsoft Fabric that investigates datasets, executes PySpark, and validates results. Persistent, shared tasks retain project context while teams define operating boundaries, review evidence, and guide changes. Tasks can be initiated through GitHub Copilot CLI, Claude Code, or Codex.

Daniel Roth outlines Microsoft’s collaboration with Syncfusion on Blazor, including a free, MIT-licensed component toolkit and merged framework and scaffolding fixes. He also covers testing contributions for .NET 11 and proposed setup tooling and AI assistant skills, distinguishing available improvements from contributions still under review.

LeeBenjamin introduces the generally available Govern experience in Microsoft Fabric’s OneLake catalog and the new Govern Skills collection. Together, they provide centralized estate insights, policy controls, and AI-assisted governance checks, with evidence-based recommendations and confirmation before sensitive changes through tools such as GitHub Copilot.

The Visual Studio Code Team outlines changes in VS Code 1.145 Insiders, including improved worktree cleanup, experimental previews of web apps on remote agent hosts, and new agent-session controls. The updates also address keyboard accessibility for MCP authentication prompts and add symbol breadcrumbs for unsaved files.

Satya Nadella outlines GitHub updates announced at Universe, including Copilot Fusion performance and cost claims, a unified Copilot harness in Foundry, and Copilot Projects for coordinating multiple agents across repositories. The post also notes plans to expand GitHub capacity and strengthen its architecture.

Jeff Hollan, Viswajeeet Balaji, and Anthony Chu introduce the Microsoft Copilot harness public preview in Foundry Agent Service and generally available resilience for hosted agents. They demonstrate reusable skills and declarative deployment, explain recovery responsibilities and preview limitations, and outline consumption-based runtime billing beginning December 1, 2026.

Jay Parikh outlines GitHub’s Universe 2026 announcements, including multi-repository Copilot projects, model orchestration, deeper code review, and enterprise access billed through AI credits. He also describes GitHub’s infrastructure expansion, redesigned Git architecture, and ongoing migration to a multi-region Azure deployment.

GitHub Staff examines the 2026 Octoverse findings on repository activity, agent-started pull requests, and changing model choices. The report connects growing use of coding agents with shifts in review workflows, GitHub Actions usage, and language participation, while explaining the telemetry’s limitations and announcing general availability of GitHub Copilot Fusion.

Allison announces the general availability of GitHub Copilot Fusion, which selects and orchestrates models for coding tasks. Developers choose between three effort levels, while administrators control eligible models and defaults. The announcement also covers supported clients, per-model billing, and a temporary discount for paid Copilot plans.

Allison introduces updates to the GitHub Copilot app, including Projects for coordinating work across repositories and connected tools, and computer use in public preview on macOS and Windows. The app can return pull requests for review, work with design context, and control desktop apps with user approval.

Allison outlines a new GitHub Enterprise licensing model that includes GitHub Copilot access for licensed users, with AI usage charged against an enterprise budget. The announcement explains which features remain available without budget, the $10 monthly standalone option, and initial availability for enterprises working with GitHub or Microsoft account teams.

Allison introduces the GitHub security extension, now in public preview for the GitHub Copilot app. Developers can browse code scanning alerts, ask Copilot to explain or fix them locally, and review changes before opening pull requests. The announcement also covers permissions, AI credit usage, and supported GitHub environments.

Allison introduces GitHub intelligence, a context layer combining persistent Copilot memories, opt-in dreaming, and cross-repository code search. Copilot Memory is generally available on paid plans, while dreaming enters public preview with controls for personal and repository context, pull-request review of instruction changes, and AI-credit costs.

Allison introduces shared Copilot Automations, which let teams maintain repository-owned AI workflows for bug reproduction, pull request screenshots, and incident investigation. The rollout adds shared visibility, editing permissions, repository-owner billing, and workflows that keep running when their creator leaves, with an administrator policy required for organization-owned repositories.

Allison announces general availability of repository plugin defaults and enterprise-managed plugin controls for GitHub Copilot. Maintainers can configure default plugins, while enterprise administrators can require or block plugins and restrict marketplaces. The controls work across VS Code, Copilot CLI, and the GitHub Copilot app, with managed settings overriding personal choices.

Allison outlines GitHub Copilot updates in VS Code, including cloud session continuity, persistent local and remote agents, and mobile controls. The releases also cover parallel model comparisons, experimental desktop interaction, custom clients through Agent Host Protocol, and preview marketplaces and connectors with enterprise policy controls.

Allison introduces image generation and editing in GitHub Copilot, powered by GPT Image 2.5 Flare. The public preview lets developers request images conversationally, preview them inline, and save them to their workspace. Image tokens are billed separately from chat, while administrators can disable the feature through model policies.

Allison announces an opt-in experimental preview of GitHub Copilot Fusion routing between supported local and cloud models. The update covers managed Windows setup and manually configured CLI models across Windows, macOS, and Linux, with important limits: routing still requires the cloud, and local models may offer different safety protections.

Allison announces general availability of GitHub Agentic Workflows, which compiles natural-language Markdown into GitHub Actions workflows powered by coding agents. The release adds runner support, model routing, and stronger isolation, with read-only defaults and validated outputs. The default GitHub Copilot CLI engine requires a Copilot license.

Sandeep Sen summarizes October's Azure SDK releases, including authentication changes in Azure Core for .NET, managed identity improvements for Python and Java, and the first stable Cosmos DB libraries for Rust. The update also lists new stable management libraries and initial previews for AI and provisioning libraries.

Hailey Victory reviews how VS Code has evolved since GitHub Universe 2025, covering parallel agent sessions, shared cloud workflows, multiple harnesses and models, and project customization. The overview explains how developers can coordinate agent work across tools while retaining control over editing, debugging, access policies, and reviewing changes.

Allison announces public preview support for server-to-server authentication in GitHub Copilot. GitHub Apps can use short-lived installation tokens to make SDK requests on behalf of an organization without user credentials. Usage is billed to the organization, with its model policies, budgets, spending controls, and app access controls applied.

Mark Downie outlines the October Visual Studio 2026 update, including comparisons between historical commits and working trees, bulk merge conflict resolution, and a refreshed GitHub Copilot Chat experience. The update also adds opt-in preservation of unsaved work, keyboard-based tab reordering, and searchable Fonts and Colors settings with inline previews.

Daniel Roth outlines Infragistics’ contributions to Blazor, including more than 50 free, MIT-licensed controls and new analyzers merged into ASP.NET Core for .NET 11. He also covers component setup requirements, a scaffolder under review, and a .NET skill that helps AI coding assistants configure Ignite UI applications.

Allison announces general availability of assisted permissions in GitHub Copilot CLI, GitHub Copilot app, and Visual Studio Code. A language model evaluates command, file-path, and URL requests, automatically allowing those classified as clearly safe. Enterprise administrators can control available permission modes and set defaults for new sessions.

Alper explains how Microsoft Fabric’s preview React visualization libraries support analytics apps built with AI coding agents. Using Vega-Lite specifications and shared data, theming, and interaction contracts, the components provide reusable charts and grids with formatting, scrolling, tooltips, and consistent styling rather than requiring each app to recreate these behaviors.

Allison announces GPT-6.1 Terra’s general availability in GitHub Copilot for everyday agentic coding and terminal tasks. The model uses provider list pricing under usage-based billing, with gradual rollout across supported clients. Business and Enterprise administrators can control access through model policies and default enablement settings.

sbaynes describes how Microsoft combines AI agents, testing, static analysis, and formal verification to transform large codebases. The article covers Windows toolchain migration, C/C++ modernization into Rust, and specification-driven validation, explaining how small expert teams review shared knowledge and system properties rather than every generated change.

sbaynes describes using the Microsoft Discovery app to adapt insect-monitoring systems for the Asian citrus psyllid, a pest that spreads citrus greening disease. The case study covers grounded AI agents, experimental design, permitting, and synthetic biometric data for preliminary edge classifiers, with scientists reviewing decisions throughout the workflow.

Rena Liu introduces Azure AI Speech LLM 09-2026, adding 25 languages and faster multilingual transcription. The announcement covers recognition improvements, locale selection, phrase lists, and REST API configuration, with automatic service-side deployment requiring no customer action.

Cynthia Zanoni and Pooja Yarabothu explain how Microsoft’s PostgreSQL extension brings database exploration, SQL editing, and troubleshooting into VS Code. They show how GitHub Copilot can help interpret schemas and query plans, while emphasizing result verification and careful review of performance recommendations before making changes.

Ben Steenhoek and Ulugbek Abdullaev explain how configurable eagerness lets paid GitHub Copilot subscribers adjust inline suggestions in VS Code. They describe the low, medium, and high modes, the training approach behind them, reward-hacking mitigations, and experimental results showing how suggestion frequency and size change.

Allison announces an update to the GitHub Copilot usage metrics API that adds a new totals_by_client breakdown, letting enterprises and organizations see request, session, and token usage coming from tools outside the IDE (like OpenCode, Copilot CLI, VS Code, and JetBrains).

Allison announces an updated GitHub app for Microsoft Teams that combines GitHub notifications with GitHub Copilot-powered actions, letting teams respond to issue and pull request updates by mentioning @GitHub to investigate, fix bugs, run tests, and open pull requests from within Teams.

Microsoft Security Research, Mahesh Mandava, Srinivasan Govindarajan, Sagar Patil and Suriyaraj Natarajan break down the LiteLLM 1.107.0 supply-chain compromise, showing how Python startup hooks enabled credential theft without importing LiteLLM, what data was targeted, and how to investigate and recover using Microsoft Defender detections and hunting queries.

sbaynes introduces Microsoft research on evaluating step-by-step clinical diagnostic reasoning with SDBench and the Microsoft AI Diagnostic Orchestrator (MAI-DxO), including accuracy and cost results across NEJM and MIMIC-IV-ED cases and plans to open-source parts of the benchmark for reproducible evaluation.