Daily AI News Brief - July 24, 2025

July 24, 2025 presents ten significant AI developments spanning real-time translation, search API services, design automation, voice synthesis, video generation competition, infrastructure expansion, photo enhancement, content creation tools, historical research applications, and web development democratization.

1️⃣ ByteDance Releases End-to-End Simultaneous Interpretation Model Seed LiveInterpret 2.0

ByteDance Seed team has launched their latest research achievement - Seed LiveInterpret 2.0, a model that achieves industry-leading levels in Chinese-English simultaneous interpretation quality, featuring low latency and real-time voice cloning capabilities that significantly enhance the naturalness and fluency of cross-language communication.

Available with technical details at https://arxiv.org/pdf/2507.17527, the model represents a breakthrough in simultaneous interpretation technology, enabling natural multilingual communication through advanced speech-to-speech generation with voice preservation capabilities.

2️⃣ Metaso Search API Launches: 3 Cent Pricing Provides Multimodal Search Capabilities

Metaso AI Search has officially launched its Search API, providing developers with a new alternative to Bing Search API. The API is priced at 0.03 yuan per query, supports multimodal search, and has no usage barriers for quick integration and immediate testing.

The service is based on Metaso's self-built multi-language index database of hundreds of billions scale developed over the past year, which has been tested with tens of millions of daily calls in the Metaso AI Search product, supporting web, image, video, and document multimodal search.

3️⃣ Lovart AI Official Version Global Launch: Full-Chain Intelligent Design Reshapes Creative Experience

Lovart AI official version global launch emphasizes its innovation as the first artificial intelligence design Agent, redefining design industry standards through natural language interaction and full-chain design capabilities. The article highlights new functionality ChatCanvas and the China-market-focused Xingliu Agent, demonstrating profound industry impact.

The platform enables users to generate up to 40 professional design products within minutes from simple text descriptions, covering brand visuals, marketing posters, and short videos with quality comparable to top advertising agencies through its proprietary Creative Reasoning Engine.

4️⃣ Li Mu Team Releases Higgs Audio v2, Opening New Era of Voice Synthesis

Li Mu team's Higgs Audio v2 represents a major breakthrough in voice synthesis, featuring multilingual dialogue generation, automatic prosody adjustment, and voice cloning capabilities. The model incorporates 10 million hours of voice data training and demonstrates outstanding performance across multiple tests, becoming an industry benchmark.

The model demonstrates leading performance in generating lifelike and emotionally competent voice with over 75% win rate against ChatGPT 4o in benchmark evaluations, trained on over 10 million hours of audio data with sophisticated processing and annotation pipeline.

5️⃣ Sora2 Surfaces: OpenAI Aims to Regain Leadership in Generative AI Video Field

OpenAI is developing Sora2, the successor to its text-to-video model Sora, while Google Veo3 gains widespread adoption. This indicates increasingly fierce competition in the generative AI video field as companies vie for market leadership in video generation technology.

The competitive landscape reflects the strategic importance of video generation technology, with both companies investing heavily in next-generation models to capture market share in the rapidly expanding AI video creation market.

6️⃣ OpenAI and Oracle Partner to Expand Stargate Project, Creating Thousands of Jobs

OpenAI has reached a new agreement with Oracle to expand Stargate project data center capacity in the US to 4.5 gigawatts, with total capacity exceeding 5 gigawatts. This marks an important step toward OpenAI's goal of achieving 10 gigawatts by 2029, aimed at making the US the dominant force in global AI development.

The massive infrastructure investment represents one of the largest AI development initiatives globally, positioning the United States as a leader in AI computational infrastructure while creating significant employment opportunities across technology sectors.

7️⃣ Google Photos Adds AI Features: Photos Become Anime in Seconds, One-Click Video Generation

Google Photos has launched multiple AI-based new features including converting static photos into dynamic videos and transforming photos into different artistic styles. These features aim to enhance user creative experiences through experimental approaches for continuous product optimization.

The new features leverage advanced AI models to democratize creative content generation, enabling users to transform everyday photos into engaging multimedia content with professional-quality results through simple interface interactions.

8️⃣ YouTube Shorts Launches New AI Effects: Photos Become Videos in Seconds

YouTube has announced revolutionary generative AI features for Shorts creators including image-to-video and AI effects. These tools can transform static photos into dynamic videos with various creative options, significantly lowering creation barriers while enhancing content attractiveness.

The platform's AI integration represents YouTube's commitment to empowering creators with advanced tools, making professional-quality video production accessible to content creators regardless of technical expertise or resource constraints.

9️⃣ Google Launches Aeneas Model: Opening New Paths for Ancient Text Interpretation

Google's Aeneas model provides new methods for ancient inscription interpretation, accelerating historians' work in inscription restoration, identification, and dating through artificial intelligence technology. The model can extend to other ancient languages and materials, greatly improving historical research efficiency and depth.

Available with details at https://deepmind.google/discover/blog/aeneas-transforms-how-historians-connect-the-past/, the model represents breakthrough application of AI in historical research, enabling scholars to process and analyze ancient texts at unprecedented scale and accuracy.

🔟 GitHub Spark Emerges: One Sentence Creates Web Applications, AI Development Enters New Era

GitHub Spark uses natural language processing technology enabling both developers and non-developers to quickly build personalized web applications, significantly lowering programming barriers and providing new possibilities for micro-application development through intuitive creation processes.

The platform represents a paradigm shift toward natural language programming, making web application development accessible to users without traditional coding skills while maintaining the flexibility and power needed for meaningful application creation and deployment.