
Microsoft patents AI technology to modify speech and faces in live video
Microsoft has filed a patent for an AI system capable of altering video and audio in real-time, including lip movements and word replacement.
Category
AI models, automation and artificial intelligence news.
1656 articles

Microsoft has filed a patent for an AI system capable of altering video and audio in real-time, including lip movements and word replacement.

Anthropic's public offering reveals deep ties with Amazon, Google, Microsoft, and Broadcom as investors, suppliers, and rivals.

Anthropic has started massively restricting paying Claude users over attempts to automate and fine-tune large language models via API.

Google is retiring Gems in Gemini and replacing them with Skills. Learn about the new feature and how to migrate your custom prompts from Claude.

French AI company Mistral has released open weights for its trillion-parameter Mistral Large 4 model, featuring three reasoning modes and high token efficiency.

Study of six AI models reveals a tendency to recommend unnecessary medical tests. In ambiguous cases, errors reached 80%.

DeepSeek has unveiled Super-Agent, an open-source AI prototype designed to read developer chats and suggest code solutions autonomously.

University of Kentucky administrator David Darman secured religious exemption from using AI. The university approved his request in September 2026.

Learn how to use a shared workspace folder and instruction files to seamlessly switch between AI coding agents without losing context.

Developers are using AI to reverse-engineer and clone Adobe Creative Cloud apps, sparking intense debates over piracy and software monopolies.

OpenAI CEO Sam Altman met with Samsung executives in South Korea to address growing memory chip shortages and future AI data center infrastructure.

OpenAI has announced it is doubling Pro subscription limits and rolling out ChatGPT Work capabilities directly into regular chat.

Former OpenAI co-founder and Tesla AI chief Andrey Karpathy built HackerTube in two hours, featuring only videos recommended on Hacker News.

Neuralink's second product, Blindsight, is fully ready from an engineering standpoint and is currently awaiting regulatory approval for human trials.

Google Photos has integrated Gemini Task to enable autonomous photo management, batch organization, and AI-powered image editing.

Google's Suncatcher space data center prototype has established communication with Earth and successfully tested the Gemma 4 AI model in orbit.

An experiment with Gemini Spark on Mac demonstrates how AI can intelligently organize messy folders by project and context rather than just file types.

Scientists successfully reconstructed the origin history of 1.28 million mouse embryo cells using the DNA Typewriter molecular recorder.

Anthropic and OpenAI have presented California authorities with contrasting approaches to regulating and emergency-stopping potentially dangerous AI models.

Microsoft and Google are adopting a unified IETF standard allowing website owners to block AI training and content generation without losing search traffic.

Apple Intelligence is launching in China this December with iOS 27.3. Some features will be powered by Alibaba's Qwen AI model.

AMD has deployed its first Helios server platforms in Meta's data centers for testing and plans to invest $1 billion in partner shares to boost AI capabilities.

Anthropic shared results of Claude working with six biotech companies, generating 35,000 antibody variants and speeding up documentation reviews by 89%.

Neuralink linked its N1 brain chip to Grok AI. A speech-impaired patient managed to control AI and make phone calls using thoughts.