AI model Archives

AppWizard

September 6, 2026

I spent months testing the top Android keyboards; one easily replaced Gboard for good

The author initially favored Google Gboard but began exploring alternatives due to declining typing accuracy and clipboard management issues. They found Microsoft SwiftKey's stable version effective for swiping, predictive text, and multi-language support, but were disappointed with its voice input feature. Transitioning to SwiftKey Beta improved voice typing significantly, though it still lacks support for some niche languages and font scaling options. FUTO Keyboard is noted for its privacy and offline functionality but has limited language support and clipboard management. Samsung Keyboard, now the author's primary choice on their Galaxy Tab, offers deep integration with One UI and improved text prediction, though voice input still needs work. Gboard remains competent, especially in voice typing, but the author feels it is becoming less appealing due to data harvesting concerns and limited customization. Currently, the author plans to continue using Microsoft SwiftKey Beta, which meets most of their typing needs.

AppWizard

September 3, 2026

RPCS3 emulator devs slam Nvidia DLSS 5 as ‘AI-slop generator’ — says industry pushing ‘more upscalers and frame generation to hallucinate games and hide their lack of optimisation’

Developers at RPCS3 have criticized Nvidia's DLSS 5 technology, calling it an “AI-slop generator.” They noted that some individuals are testing a leaked DLSS 5 DLL on the RPCS3 renderer. The official RPCS3 account stated that they do not support technologies that obscure optimization issues. DLSS 5 is set to officially launch with NBA2K27 and aims to provide advanced neural rendering for improved visual realism. It will feature various AI models and controls for artists, but is resource-intensive and compatible only with RTX 50-series GPUs and GeForce Now. Some modders have managed to use the leaked DLL on older RTX 40-series graphics cards.

TrendTechie

September 2, 2026

Sony и Warner подают в суд на Anthropic из‑за обучения Claude на пиратских песенниках

Sony Music Publishing and Warner Chappell Music have filed a lawsuit against Anthropic, claiming that the company used pirated songbooks and song lyrics from "shadow libraries" to train its AI model, Claude. The lawsuit cites approximately 7 million books obtained through torrent downloads and web scraping, including unauthorized sheet music and songbooks. The plaintiffs are seeking statutory damages of up to 0,000 for each work that is found to have been willfully infringed. The lawsuit was filed on August 28, 2026, in the U.S. District Court for the Northern District of California, naming Anthropic and its co-founders as defendants. The plaintiffs allege that the data sources used for training included 5 million books downloaded from Library Genesis and 2 million from Pirate Library Mirror, as well as scraped song lyrics from licensed platforms Musixmatch and LyricFind. The use of BitTorrent for downloads introduces legal risks due to the simultaneous distribution of files. The lawsuit emphasizes the distinction between legally purchased and pirated materials, with prior cases indicating that downloads from pirate libraries constitute significant copyright infringement.

AppWizard

August 11, 2026

Even the AI assistant inside Russia’s state-backed Max messenger tells users not to trust the app

Alisa, the AI assistant in the Max messenger, has recommended users consider alternative messaging platforms such as Telegram, Signal, or WhatsApp. She highlighted several security vulnerabilities in Max, including the absence of end-to-end encryption and the direction of user data to VK-owned servers. The data collected by Max includes IP addresses, activity timestamps, contact lists, device and browser types, internet service provider information, and various metadata. Alisa suggests that users keep Max on a separate device, referred to as a “maxphone,” to protect sensitive information. If used on personal devices, she recommends using the browser version and restricting the app's access to location services, contacts, microphone, and camera to enhance privacy and security.

Winsage

August 8, 2026

Google Chrome, Microsoft Edge could quietly download up to 20GB AI models on Windows 11

Microsoft Edge and Google Chrome are phasing out support for Manifest V2 extensions, affecting popular add-ons like uBlock Origin. Google Chrome can automatically download a 4GB AI model for local functionalities, requiring approximately 20GB of free disk space for the download. Microsoft Edge has established a similar 20GB minimum storage requirement for downloading local AI models, with additional specifications including Windows 10 or 11 and macOS 13.3 or later, at least 5.5GB of VRAM, and an unmetered network connection. Users can disable on-device AI features in Chrome, and Edge will automatically delete the AI model if available storage drops below 10GB.

AppWizard

August 4, 2026

Google Merges AI Mode, Search Live, and Song Search Into Android Voice Button

The four-color microphone button on Android homescreens has been updated to provide four distinct functionalities: Search, Translate, Live, and Create. This redesign includes a mode switcher similar to Google Lens, allowing users to select their preferred mode instead of relying on the app to infer intent. Each mode is powered by different AI architectures: AI Mode uses Gemini 3.5 Flash for complex queries, Search Live employs Gemini 3.1 Flash Live for real-time conversations, and Song Search identifies melodies using machine learning. The toolbar now remembers the last selected mode, enhancing user convenience. The new features are currently available in beta versions of the Google app (17.1 and 17.2) on Android, with no specified timeline for broader rollout.

Winsage

July 30, 2026

Microsoft Quietly Adds New Windows App That Wants to Scan Your Face

Microsoft has started rolling out a new application called OneDrive Photos to select Windows 11 PCs, which automatically indexes local photo libraries and enhances user experience with AI-driven search capabilities and optional facial grouping features. The app can identify and showcase images stored locally, even without a Microsoft account, and provides AI-powered search functionality for locating photos through natural-language queries and optical character recognition. It includes a People feature that groups similar faces while ensuring privacy, with the option to disable it. OneDrive Photos is integrated with the existing OneDrive client and cannot be uninstalled independently; users must uninstall OneDrive entirely to remove the app. Microsoft has not commented on the rollout.

Winsage

July 29, 2026

Windows 11’s August Patch Tuesday update brings several important changes, and I break down the biggest ones

Microsoft will release the Windows 11 August 2026 Security Update on August 11, 2026, featuring new functionalities and enhancements. Key additions include Windows Hello Enhanced Sign-in Security, the ability to uninstall AI models from Copilot+ PCs, new touchpad gestures, voice isolation for Voice Access, Fluid Dictation for Voice Typing, and a redesigned Start menu with parental controls notice. File Explorer improvements include displaying file sizes in appropriate units, middle-clicking to open folders in new tabs, and fixes for bugs. Users can manage AI models through Settings, and the Lock screen will default to the Weather widget. Touchpad settings will allow customization of scroll and zoom speeds. Fluid Dictation will be deactivated by default, with tooltips for guidance, while Voice Access will feature enhanced speech recognition settings. The Account Control interface will be redesigned for clarity, and power settings will apply across all plans. Windows Update will provide better progress estimates and improved post-update cleanup. Enhancements to system sounds in dark mode will also be included.

AppWizard

July 24, 2026

I completely reimagined my videos with Video Remix in Google Photos — here’s how you can, too

Google has introduced an AI model called Gemini Omni in the Create tab of the Google Photos app, featuring a tool named Video Remix. This allows users to transform 10-second video clips into new creations using various templates. The feature enhances existing content but has limitations, such as not generating remixes for videos with faces of minors. For creating new video content, the Omni model in the Gemini app is preferred, while Video Remix is more convenient for minor modifications. Users can access Video Remix through a subscription to Google AI, and purchasing a Google Pixel 10 Pro includes a complimentary year of Google AI Pro.

Winsage

July 21, 2026

Windows 11 will let you search for your files in plain language

Microsoft is testing a new semantic search feature for Windows 11's File Explorer, which aims to improve file search by understanding user intent rather than relying on exact word matches. This feature allows users to search in natural language, such as "photos from last summer" or "Europe trip budget," and can identify files based on context, even if specific names are not mentioned. The functionality is powered by an AI model that operates locally on users' machines to maintain privacy and does not require an internet connection. However, it is only available on Copilot+ PCs, which have specialized AI chips capable of executing at least 40 trillion operations per second, limiting access for users with older machines. Currently, the feature is available only to Windows Insiders as Microsoft rolls it out in test builds.