The Witcher 3 uses gen-AI to enhance Geralt's Polish voice following actor's stroke - "I do think this is one of those ways where you can use AI ethically"

eurogamer.net Nov 12, 2026

The Witcher 3 uses gen-AI to enhance Geralt’s Polish voice following actor’s stroke – “I do think this is one of those ways where you can use AI ethically”

Curated by Shubham · TechCapsules · Nov 12, 2026

AI-summarised brief · reviewed before publication

CD Projekt Red has used generative‑AI voice‑altering technology to restore Polish actor Jacek Rozenek’s Geralt voice for the upcoming Witcher 3 expansion Songs of the Past. After a 2019 stroke left Rozenek’s voice deeper and rougher, the studio, with his consent, partnered with Respeecher to train an AI on his earlier recordings, producing lines that match the original 2015 sound. The company has applied similar AI techniques in Cyberpunk 2077, citing ethical use and consent. The expansion, slated for release next year, will not be available on older consoles.

💡 Why It Matters

· The move shows how AI can preserve a character’s iconic voice after an actor’s health changes, ensuring continuity for fans.
· It also sets a precedent for ethical voice synthesis in gaming.

AI Capsules Gaming

Read full article on eurogamer.net

Runway’s Praxis-1 world action model for robotics is now available

Runway AI Inc. released Praxis‑1, a 27‑billion‑parameter world‑action model for robotics, now available under an Apache 2.0 license. The model, built on Runway’s video foundation, predicts video and actions simultaneously and has achieved top rankings on NVIDIA’s RoboLab leaderboard. Partners such as Noble Machines, Standard Bots, and Ultra have deployed Praxis‑1 on diverse hardware—from legged humanoids to six‑axis arms and dual‑arm warehouse robots—demonstrating rapid learning from limited examples. The open‑weight release includes full fine‑tuning and LoRA support for custom embodiments.

💡 By providing a single, adaptable policy that works across multiple robot types, Praxis‑1 reduces the need for task‑specific training, accelerating deployment in industrial settings.

ign.com · Nov 12

Pokémon Fan Who Spent Thousands of Dollars on AI Tools Upset as…

A Pokémon enthusiast, Big Kid Nation, has voiced frustration after Anthropic’s Claude AI models began refusing to generate code involving official Pokémon assets. The fan, who has reportedly spent up to $2,000 monthly on “vibe‑coded” AI tools, claims no revenue has been earned from his project, Pokémon Arena 2D, despite significant investment. The game, a free 2D pixel‑art arena, has avoided legal scrutiny so far but may face increased pressure as Anthropic tightens copyright restrictions.

💡 The crackdown illustrates how AI‑powered development can clash with intellectual‑property rights, potentially stifling fan‑made innovation and altering the landscape for independent creators.

techcrunch.com · Nov 12

Five ways Jensen Huang’s ouster from Nvidia could damage the AI industry

Nvidia’s board removed CEO Jensen Huang on Thursday after he admitted to an affair with a woman suspected by U.S. authorities of being a Chinese spy and an executive of a Singapore‑based cloud firm accused of funneling Nvidia chips to China. The scandal intertwines personal misconduct with national‑security concerns, prompting congressional committees to seek Huang’s testimony and sparking investigations into his political donations, including a $25 million contribution to a Trump‑linked project. Huang’s ouster also strips the AI sector of its most vocal champion for open‑weight models, self‑regulation, and the view that AI is merely an engineered tool. His departure jeopardizes Nvidia’s pending $12.9 billion acquisition of Hugging Face and leaves a leadership vacuum in industry‑wide safety initiatives.

💡 Without Huang’s influence, the push for open‑weight AI and coordinated self‑regulation loses its most forceful advocate, potentially slowing collaborative security efforts across the sector.

windowscentral.com · Nov 12

Microsoft wants developers spending less time cleaning up AI-generated code

Microsoft announced a new set of developer tools aimed at reducing the time programmers spend fixing code generated by artificial‑intelligence assistants. The suite, integrated into Visual Studio and GitHub Copilot, adds automated linting, context‑aware refactoring and security‑focused scans that run immediately after AI‑produced snippets are inserted. Microsoft says the features will identify syntactic errors, style violations and common vulnerability patterns, offering one‑click corrections while preserving the original intent of the generated code. The company demonstrated the workflow on a sample project, showing a 30 percent drop in manual edits compared with baseline Copilot usage. The initiative reflects Microsoft’s broader push to make AI‑assisted development more reliable and to address developer concerns about the quality and safety of machine‑written code.

💡 By automating cleanup, Microsoft aims to turn AI code suggestions from a time‑sink into a productivity boost, encouraging wider adoption of AI‑assisted programming.

9to5google.com · Nov 12

Google rolling out Gemini app for CarPlay

Google has launched the Gemini app on CarPlay, enabling iOS drivers to interact with Gemini Live while on the road. The interface features a waveform indicator that shows listening, thinking, and responding states, with “End” and “Mute” controls. Users can interrupt responses, pause, or add details, and conversations sync with the iPhone Gemini app, allowing continuity across devices. The rollout is global, supporting all Gemini Live languages, and requires the latest Gemini app update. The feature aims to serve as a conversational driving companion that reduces distraction.

💡 By extending Gemini to CarPlay, Google expands its AI ecosystem into a high‑stakes, safety‑critical environment, offering a unified conversational experience across Android and iOS platforms.

9to5mac.com · Nov 12

iCloud+ has new perk in iOS 27 if you pay for more…

Apple has added a new AI feature to its iCloud+ service in iOS 27, targeting users on the higher‑tier storage plans. The feature, called Apple Intelligence, automatically generates video descriptions for HomeKit Secure Video recordings, enabling natural‑language searches within the Home app. It is available only on the $9.99, $29.99, and $59.99 monthly plans and requires a compatible HomeKit camera, a home hub such as HomePod or Apple TV 4K, and an iPhone that supports Apple Intelligence. The update expands iCloud+ beyond storage, offering added value for users who already need large cloud capacity.

💡 By bundling AI‑driven video search with premium storage, Apple incentivizes higher‑tier subscriptions and strengthens its ecosystem lock‑in. The feature demonstrates Apple’s push to integrate machine learning into everyday home security workflows.