Microsoft is previewing the Microsoft Bing AI Search Choice extension for Google Chrome and Microsoft Edge. It provides a global switch for enabling or disabling Bing’s AI-generated answers, while the “-ai” query operator can return traditional results without an extension; installing the extension also makes Bing the browser’s default search engine.
Adobe announced that Adobe-powered AI editing and Adobe Express templates are coming to the Windows 11 Photos app, with a preview for Windows Insiders scheduled to begin in December 2026. The additions include extending images, filling empty edges created by straightening photos, changing selected content, removing distractions, and previewing personalized templates before continuing in Adobe Express. Adobe’s announcement does not specify compatible PCs, an Insider channel, an exact preview date, a public rollout schedule, pricing, account requirements, or generative-credit limits. The parties also have not clarified which templates will be free or whether trial prompts will appear.
Edelbyte emerged from stealth with $70 million raised across a $25 million seed round led by Sequoia and a $45 million Series A led by Quiet Capital, according to Forbes and 科技新报. The Tel Aviv- and San Francisco-based startup was founded by four former Microsoft engineers and is building search and analytics infrastructure for AI agents. 科技新报 reported a $500 million valuation. Edelbyte says its platform can combine analytics, text search, spatial queries, and AI workloads while supporting public-cloud, private-cloud, and on-premises deployment.
Charles Humble argued at GOTO Copenhagen that sustainability should be a top-level concern in AI architecture, alongside cost, performance, security, compliance and reliability. He recommended using prebuilt or smaller models, training with renewable energy, and placing or shifting workloads when lower-carbon power is available. InfoQ also cited IEA estimates and company reports showing rising electricity use and emissions, including Microsoft’s and Google’s attribution of much of their growth to AI infrastructure.
TechNews reports that Amazon shortened the estimated useful life of some servers and networking equipment from six years to five in 2025, citing faster AI and machine-learning progress. The change increased depreciation and amortization by $1.4 billion and reduced net income by $1 billion, with AWS identified as the business most affected. TechNews also cites a forecast that Amazon, Microsoft, Alphabet and Meta will spend a combined $725 billion on capital expenditures in 2026.
Microsoft’s Automatic Super Resolution (Auto SR) has exited preview and is now available for ROG Xbox Ally X, ROG Xbox Ally X20, and selected AMD Ryzen AI-powered Windows Copilot+ PCs. The update adds handheld and battery operation, 1080p input support, a Performance Mode with a reported 120 FPS ceiling, and compatibility with Windows Auto HDR, though some features are rolling out gradually. Microsoft’s Rocket League demonstration showed an average of about 70 FPS in Quality Mode and 115 FPS in Performance Mode on ROG Xbox Ally X at 720p with high graphics settings; Wccftech notes that these are vendor figures rather than independently verified benchmarks.
Microsoft released .NET 11 at .NET Conf 2026. The release adds C# 15 union types and closed hierarchies, along with updates to ASP.NET Core and .NET MAUI. The SDK uses a Native AOT command-line host by default and keeps the MSBuild server running between commands. Microsoft also reports hundreds of runtime optimizations and lower everyday command overhead. AI-related additions include an MCP server template, Microsoft.Extensions.AI, Microsoft Agent Framework, experimental Blazor AI components, .NET Skills, and .NET MAUI DevFlow. ASP.NET Core also gains asynchronous validation and OpenAPI 3.2 support.
Anthropic began its IPO roadshow and said in an amended SEC prospectus that it plans to list on Nasdaq under the ticker “ANT,” with a proposed price range of $70–$78 per share. The report says the maximum valuation is about $1.4 trillion on a fully diluted basis, while the valuation excluding potential shares such as options and restricted stock is about $1 trillion. The offering would comprise 825 million Class A shares, including about 801 million newly issued shares and 23.65 million shares offered by existing holders. Underwriters could buy up to about 124 million additional shares from Anthropic within 30 days. The prospectus also proposes a $26 billion private placement of Class N common stock by companies affiliated with Google, Nvidia, AMD, Broadcom, Micron, Microsoft and TSMC, contingent on completion of the IPO and other agreed conditions.
Microsoft researchers demonstrated Lazarus, a technique that reprograms fewer than 10 dormant neurons to implant a trigger-activated backdoor in an LLM-based agent without retraining it. In the reported demonstration, the agent behaved normally and passed standard evaluations until a trigger caused it to download and execute attacker-supplied code. Microsoft says reliable detection requires comparison with unmodified weights and recommends model provenance checks, weight auditing, permission restrictions, and monitoring.
Ohio State University and Microsoft signed a five-year partnership to bring more AI and cloud computing to Wexner Medical Center. Ohio State aims to move 70% to 80% of its storage and computing to Azure, including pathology and medical imaging, and maintain a backup of its electronic health records there. Doctors already use Microsoft Dragon Copilot to draft visit notes, which they review and sign. A platform for forecasting surgery demand and staffing remains in design, while an AI tool for cancer case reviews is only being explored. The parties did not disclose the financial terms.
Databricks announced that Unity Catalog Metric Views are now available in Power BI. The company says the integration provides a unified, governed semantic layer for reporting and AI, addressing demand for a single trusted definition of business semantics.
Proofpoint researchers described a “BluePhishing” attack that uses social engineering to persuade users to approve malicious Microsoft Entra Agent ID blueprints through a genuine authorization page. According to TechRadar, an approved blueprint can create agent identities with corporate access, with the report claiming a single blueprint can support up to 250 identities. Proofpoint recommends restricting third-party agent consent, applying least-privilege controls, and monitoring agent-creation events.
Matthew Berman demonstrated a playable Gears of War: E-Day-inspired third-person shooter built in Unreal Engine 5 with Claude Fable 5.2 after providing a three-line prompt and enabling Ultra Code mode. The project reportedly ran for 43 hours and 48 minutes, consumed 8.63 billion tokens and had an estimated API-equivalent cost of $14,710, while the footage showed cover mechanics, active reloading, enemy AI and allied revivals. The demonstration included awkward vaulting animations and was not independently assessed for performance, stability or maintainability.
The Frontier Model Forum released version 1.0 of its Jailbreak Rating Framework, a proposal modeled loosely on CVSS that rates LLM jailbreaks by capability gain, capability scope, deployment ease, and discovery ease. Tom’s Hardware reports that the framework combines severity and likelihood into a rating from low to critical, with the capability-gain score constraining the other dimensions. FMF says the cross-vendor scale could help teams developing patches, but version 1.0 remains under review, and the group advises against relying on it while comments and suggestions are collected through December 21.
36Kr reports that AI companies are turning agent runtime harnesses into hosted services. Microsoft expanded Copilot Harness across its products and offered managed services through the cloud, while Moonshot AI released Kimi Hosted Agents with a “Harness as a Service” offering. OpenAI and Anthropic had also made their internal runtime systems available to developers, while DeepSeek open-sourced its harness framework.
In a Microsoft technical white paper, Microsoft reported that keeping Claude Opus 4.8 unchanged but switching the harness behind Copilot Studio increased the complex-task pass rate by 22.1 percentage points. The new system nearly doubled model calls per task, yet reduced calculated input-token consumption by 21%.
Anthropic launched its IPO roadshow, proposing 825 million Class A shares at $70–$78 each, with 801.35 million offered by the company and about 23.65 million sold by existing shareholders. It also filed for a separate $26 billion private placement involving Google, Nvidia, AMD, Broadcom, Micron, Microsoft, and TSMC Partners. The Next Web reported that the offering was oversubscribed, while Financial Times reporting indicated bankers were considering a valuation of $1.5 trillion or more; final pricing and the listing date remained unconfirmed.
A 智讯智库 report says AI assistants are increasingly shaping product evaluation even though direct AI checkout remains limited. Attentive found that 88% of surveyed US adults had used a chat assistant to evaluate products, while only 10% of final purchasers completed a purchase inside AI. The report also examines Doubao’s new Douyin Life Services fees and estimates that, under its baseline scenario, AI-influenced decisions and AI-originated transactions could represent 28% and 8% of Chinese physical-goods online retail sales by 2030.
Anthropic reportedly sent banks IPO documents outlining about 866.7 million Class A shares at $70–$80 each, a maximum transaction size of roughly $69.3 billion, then withdrew the pricing details as inaccurate. The company did not explain the error or comment, while Bloomberg described the episode as a brief clerical mistake; final pricing and fundraising totals remain unsettled. The documents also showed that Alphabet, Nvidia, AMD, Broadcom, Micron, Microsoft and TSMC planned to buy $26 billion of non-voting Class N shares at the IPO price.
Microsoft released WinApp CLI 0.8 with an experimental preview of WinUI DevTools for inspecting running WinUI 3 applications, including component properties, layout, and XAML source locations. According to Microsoft, developers and AI agents can test interface changes, work with saved annotations, and query properties and source code, while the release also adds performance analysis and Visual Studio C++ project support. Interface edits are temporary and are not written to XAML source code.
Oppenheimer analyst Rick Schafer raised AMD's rating from Neutral to Outperform and set a $600 target price, representing 34% upside from AMD's reported $449.04 close on the 5th. He based the outlook on AMD's third-quarter results exceeding Wall Street expectations and forecast a 135% increase in 2027 data-center revenue. The projection also cited potential chip investment from OpenAI, Meta, Oracle, Microsoft and Humain, alongside AMD's Instinct accelerators and Helios rack-scale solution.
A published Microsoft Technology Licensing patent application, US 2026/0349820 A1, filed April 30, 2025, proposes a distributed system that divides AI training into micro-tasks and assigns them to consenting gaming PCs, laptops and smartphones based on GPU performance, memory and available time. It would monitor utilization, memory use, temperature and power consumption, then cross-validate results and weight them by reliability. The application says participants could receive credits, free software, model access or cash. Microsoft has announced no consumer service, supported GPU list, compensation standard or launch schedule, and compensation appears only as an optional implementation rather than among the 20 claims.
At the Baron Capital annual meeting on November 9, Elon Musk said digital superintelligence could expand global economic output by 20% to 30%, double white-collar productivity, and make humanoid robots as productive as five times human workers. He also predicted that 20 million to 30 million robots could match the physical output of US society and that modern currencies based on scarcity could become obsolete. TechNews attributed these figures to Musk’s forecasts. Bill Gates and Anthropic CEO Dario Amodei warned that the transition could bring severe disruption before living standards improve.
A commentary draws on The Trap by Aeer Zarah and Maurice Stuck to argue that algorithms, targeted pricing, and consumer monitoring can reshape competition. Applying that lens to large technology companies, the author questions how markets will react to heavy AI capital spending when its revenue path remains unclear. Google is presented as especially exposed because search is central to its traffic and advertising model, while the author notes that signs of users shifting from search to AI are not yet clear.
AMD and Microsoft have made Project Zenith available on the Ryzen AI Halo developer platform, initially using the Ryzen AI Max+ PRO 495 processor, according to Wccftech and IT之家 AI. It provides a preconfigured Windows 11 environment with tools including Visual Studio Code, Git, GitHub CLI, GitHub Copilot CLI, Azure CLI, Python and Node.js. AMD also added the Ryzen AI Developer Center, validated software configurations and AMD AI Playbooks. The platform supports up to 192GB of unified memory, but a corresponding 192GB retail configuration with Project Zenith preinstalled was not independently confirmed in the supplied reports.
Microsoft announced Frontier Accelerate, a cross-company benefits package spanning seven workloads: Azure, Databases, Copilot, AI-ready Productivity, Business Processes, Security, and Marketplace. It combines expert and partner assistance, pricing and project funding, cost-optimization options, and role-based training to help customers move from prototypes toward production workloads. Some benefits require eligibility, approval, and availability, while no-cost deployment assistance is subject to qualifying criteria.
IT之家 reports that Microsoft filed patent application US 2026/0341309 A1 for using AI to modify speech during live events and other real-time video programs. The described system would identify words under preset rules or viewer profiles, replace them in audio, translate speech when needed, and redraw the speaker’s lip movements to match the modified audio; processed video would reportedly be delayed by about 1 to 10 seconds. The application also lists lectures, award ceremonies, game shows, sports broadcasts, AR glasses, phones, tablets, and cameras as possible use cases.
Microsoft CEO Satya Nadella said in a June 5 episode of Reid Hoffman’s Possible podcast that he was exploring a scenario in which 200,000 employees worked with 2 million or even 20 million AI agents. He also described running 100 agents across CLI sessions, saying the resulting management burden prompted Microsoft to develop an agentic development environment. The figures were hypothetical, not an announced deployment target or current deployment scale.
A republished Chinese commentary argues that major technology-company job cuts are driven more by declining legacy businesses, cost control and organizational restructuring than by demonstrated AI substitution. It cites Google’s expansion of voluntary departures and Microsoft’s invitation to roughly 60,000 employees to leave, alongside employment, productivity and agent-performance data that it says weaken the broad claim of AI-driven mass unemployment. The commentary attributes many figures to other reports and organizations, including Reuters, the US Department of Labor and Challenger, Gray & Christmas.
Microsoft released Intelligent Terminal 0.3 with a Headless Mode that lets enabled tabs and panes continue running AI-agent tasks and compilations after the terminal window is closed. A vertical sidebar provides a consolidated view of agent status, approval requests, working directories, Git repositories, branches, and changes, while the release also detects installed agent CLIs and creates profiles for them. The experimental product is available through Microsoft Store and GitHub.
Bernstein analyst Mark Moerdler estimated that OpenAI’s disclosed compute agreements could correspond to $614 billion in infrastructure spending through 2030, below the previously cited $750 billion figure. The estimate assumes identified contracts are fully executed, while delays, cancellations, or hardware changes could reduce or postpone spending; Bernstein also expects annual spending to peak at $184 billion in 2031 as facilities enter service.
What problem it solves
Documents and office files often need to be transformed into a structured, text-friendly format before they can be processed by language models or text analysis tools. MarkItDown addresses this by preserving important document structures such as headings, lists, tables, and links during conversion to Markdown.
Key features
Broad format conversion — MarkItDown converts PDF, P…
What problem it solves
Editing text should not require extensive familiarity with terminal-based tools. The project aims to provide an accessible editor with a modern interface and input controls that are approachable for users who are largely unfamiliar with terminals.
Key features
Cross-platform installation — Installation options are documented for Windows, Linux, and macOS, including package-m…
What problem it solves
Generative AI systems and applications can contain security and safety weaknesses that require structured testing across models, endpoints, and complete applications. PyRIT addresses this with repeatable assessments, interactive exploration, and extensible workflows for red teaming these systems.
Key features
Repeatable command-line assessments — The scanner runs repeatable …
What problem it solves
Web automation and end-to-end testing often require coordinating browsers, interactions, assertions, and failure diagnostics. Playwright provides a unified API for driving Chromium, Firefox, and WebKit across tests, scripts, and AI-agent workflows.
Key features
End-to-end test runner — Playwright Test runs end-to-end tests across Chromium, Firefox, and WebKit with browser is…
What problem it solves
TypeScript adds optional types to JavaScript so tools can support large-scale JavaScript applications across browsers, hosts, and operating systems. It also compiles source into readable, standards-based JavaScript.
Key features
Optional types for JavaScript — TypeScript extends JavaScript with optional types, providing a typed language for application-scale development.
Sou…
What problem it solves
Go applications need a way to create, read, and modify spreadsheet documents produced by Microsoft Excel. They also need support for complex spreadsheet components and workflows involving large worksheet datasets.
Key features
Read and write spreadsheet files — Create and modify workbooks, set cell values, save files, and read cell values and worksheet rows through a Go API.…
What problem it solves
Knowledge workers need Claude to follow their organization’s tools, terminology, critical workflows, and preferred ways of working rather than provide only generic assistance. The repository provides customizable role-specific plugins that package this guidance with connectors, skills, commands, and sub-agents.
Key features
Role-specific plugin collection — The collection in…
What problem it solves
Products and agents need to communicate with users across multiple channels without rebuilding Inbox feeds, provider integrations, and channel webhooks. Novu addresses this with one API and a unified conversation model for product notifications and bidirectional agent conversations.
Key features
Unified multi-channel API — Send product notifications through Inbox/In-App, Pus…
What problem it solves
Applications need to run untrusted code while controlling its access to files, networks, user interfaces, and host capabilities. This requires an isolation approach that works across platforms, supports multiple containment mechanisms, and can be integrated without changing application architecture.
Key features
Cross-platform containment — MXC supports Windows, Linux, and m…
What problem it solves
Teams often have to move work into a separate AI tool, then repeatedly transfer context between that tool and the applications where collaboration happens. Switch addresses this by letting people and AI agents participate in the same conversations while keeping shared context with the work.
Key features
Shared rooms for people and agents — Each supported chat channel becomes…
What problem it solves
Speech recognition systems often lose global context when they divide continuous audio into short chunks, while speech generation must maintain speaker consistency, timing, and coherence across long recordings. The project addresses these needs with unified long-form ASR, streaming recognition, and text-to-speech models.
Key features
Long-form speech recognition — Process co…
What problem it solves
Company knowledge is distributed across business systems such as Slack, Google Drive, Jira, GitHub, and Microsoft 365. AI applications need a governed way to search this information, respect each user's access controls, and provide answers that can be traced to their sources.
Key features
Permission-aware enterprise search — Searches connected company systems while applying …
What problem it solves
Developers need support for their core edit-build-debug cycle. The project addresses this need with comprehensive code editing, navigation, and understanding support along with lightweight debugging, a rich extensibility model, and lightweight integration with existing tools.
Key features
Code editing and navigation — The editor provides comprehensive support for editing, na…
AI has already entered code writing, algorithm optimization, and model development. But after a task is completed, what remains may be reusable capability—or merely an answer and scattered logs. The difference becomes apparent only when a similar problem comes up again: can the system reuse what it learned, or must it pay the cost of exploration all over again?
Recursive Self-Improvement (RSI) co…