YouTube Enhances Shorts with AI-Powered Video Generation
YouTube is integrating Google DeepMind’s Veo AI model into Shorts, allowing creators to generate high-quality backgrounds and six-second clips. This upgrade from the previous “Dream Screen” feature enables more impressive video creation and editing capabilities. Creators can now produce standalone video clips and add filler scenes for smoother transitions. The AI-generated content will be watermarked using DeepMind’s SynthID technology. YouTube also announced new features like “Jewels” for viewer interaction during livestreams, expanded automatic dubbing support, and the ability to use AI for brainstorming video ideas, generating thumbnails, and responding to comments. The platform is also expanding its Community hubs and “hyping” feature to more channels and markets.
Runway Partners with Lionsgate for AI-Powered Film Production
Runway, an AI startup, has partnered with Lionsgate to develop a custom AI video production and editing model. This collaboration marks the first major deal between an AI video provider and a major film studio. Runway will train its model using Lionsgate’s extensive library of film and TV titles. The technology will initially be used for storyboarding, background creation, and special effects, particularly for action scenes. This partnership aims to streamline production processes and reduce costs. While AI-generated full-length films are not imminent, the technology is already assisting in producing film segments and short clips. The deal represents a significant step in AI’s integration into the film industry, potentially impacting future production methods and creative processes.
Hume Unveils EVI 2: Advanced Emotionally Intelligent Voice AI
Hume has launched Empathic Voice Interface 2 (EVI 2), an upgraded AI voice model and API. EVI 2 offers improved naturalness, emotional responsiveness, and customizability in voice interactions. The new model boasts faster response times, enhanced emotional intelligence, and customizable voices without relying on voice cloning. It supports multiple languages and can be integrated directly into apps for a seamless user experience. EVI 2 is designed to anticipate and adapt to user preferences in real-time, making it suitable for various applications from customer service to virtual assistants. With reduced pricing and increased efficiency, Hume aims to make advanced voice AI more accessible to developers and businesses, potentially revolutionizing how voice interactions are integrated into digital platforms.
LinkedIn’s AI Training Practices Raise Privacy Concerns
LinkedIn has been using user data to train AI models without initially updating its terms of service, sparking privacy concerns. The platform now offers an opt-out toggle for users in certain regions, allowing them to choose whether their data is used for training content creation AI models. LinkedIn claims to use privacy-enhancing techniques to limit personal information in training datasets. The company’s practices have drawn criticism from privacy advocates, who argue for an opt-in model instead. This situation highlights the growing trend of social media platforms repurposing user-generated content for AI training, often without explicit consent. The incident underscores the need for greater transparency and user control over personal data usage in AI development.
Apple Expands Language Support for Generative AI Feature
Apple is set to broaden the linguistic reach of its generative AI offering, the Apple Intelligence, in 2025. The expansion will include support for German, Italian, Korean, Portuguese, Vietnamese, and several English variants. Initially launching in American English with iOS 18.1, the feature will gradually roll out to other languages and regions. However, regulatory challenges are hindering its release in the European Union and China, two significant markets. Apple is actively engaging with authorities in both regions to address these issues. The company’s efforts highlight the complex interplay between technological innovation and international regulations in the AI sphere.
EzAudio: Tencent’s Breakthrough in AI-Generated Sound
Tencent AI Lab and Johns Hopkins University researchers have unveiled EzAudio, a groundbreaking text-to-audio AI model. This innovative system generates high-quality sound effects from text prompts with remarkable efficiency, outperforming existing open-source models in various evaluations. EzAudio’s novel approach, operating in the latent space of audio waveforms, eliminates the need for spectrograms and neural vocoders. While this advancement promises exciting applications in entertainment, accessibility, and virtual assistants, it also raises ethical concerns about potential misuse in deepfakes and unauthorized voice cloning. As AI audio technology progresses, balancing its potential benefits with responsible use becomes increasingly crucial.
Magnit Introduces AI Assistant Maggi to Revolutionize Workforce Management
Magnit, a global staffing and talent provider, has launched an AI-powered Integrated Workforce Management Platform featuring an assistant named Maggi. This innovative system aims to streamline the hiring process for temporary and seasonal jobs by offering a unified interface for data access, sourcing, and workforce management tools. Powered by OpenAI’s GPT-4 and connected to extensive datasets, Maggi assists with task execution, provides alerts on regulatory changes, and aids in workforce planning. The platform offers features such as automated workflows, data-driven insights, and compliance frameworks. Magnit’s goal is to simplify complex processes and improve efficiency in talent acquisition and management across various sectors.
