Adobe's new AI "Magic Fixup" uses video data for precise photo editing. It outperforms existing tech and can revolutionize industries like advertising and film. However, its ease of use raises concerns about image authenticity and ethics. Adobe's decision to open-source the technology on GitHub marks a shift in AI development, promising a new era of creativity with AI assistance.

Midjourney opens official website offering free trial with up to 25 image generation for new users. Aim to compete with xAI and Ideogram 2 by lowering entry barrier and increasing popularity. Easy registration with Google or Discord, intuitive image customization, and account merging feature for existing users.

Google released "Reimagine," an AI photo editing tool on Pixel 9. It raises ethical concerns due to its realistic effects and potential for misuse. Users can create misleading content, challenging authenticity. The tool lacks visible markers for AI-generated photos, highlighting the need for media literacy.

Microsoft has announced the Recall AI feature will be available for testing by Windows Insiders in October. Recall captures and retrieves user content, with security updates to address potential risks. The release timeline was delayed due to security concerns.

Authors are suing AI startup Anthropic for using pirated books to train chatbots. The lawsuit accuses Anthropic of profiting from stolen creativity, tarnishing the company's reputation. Anthropic defends itself by claiming fair use in US copyright law.

ByteDance's XR platform, PICO, launches first MR device, PICO 4 Ultra, with Snapdragon XR2 Gen2 platform, upgraded hardware, and software for immersive mixed reality experience. Pre-sale starts at affordable prices, aiming to compete with Apple Vision Pro. PICO continues to invest in XR field for a complete ecosystem.

OpenAI partners with Condé Nast, granting access to content archives and advanced tools. This collaboration reflects a trend of tech-media partnerships shaping the evolving content landscape. Advancements raise questions about competition, copyright, and industry adaptation.

Google releases Health Acoustic Representations (HeAR) AI system via Google Cloud API. Revolutionizing health screening, diagnosis, and monitoring using acoustic data like coughing sounds. Outperforms other models, even with limited training data. Potential for early detection of tuberculosis. Stop TB Partnership supports HeAR for tuberculosis screening. HeAR has wide applications in respiratory health research. Goal is to provide low-cost screening solutions globally. HeAR outputs low-dimension

Google is improving the Gemini writing tool in Gmail, adding a "Polish" option for users to optimize drafted emails. Shortcuts are also added for visibility on Android and iOS devices. This feature targets paid users with Gemini add-ons.

Large-scale language models (LLMs) face challenges like high resource consumption and dataset dependence. Aquila2 series models, developed by BAAI, use HeuriMentor framework for efficient training. With 100K vocab and BPE, Aquila2-34B excels in NLP tasks, showing strong competitiveness. HM framework enables real-time data optimization, accelerating model convergence and enhancing AI tech development.

Google unveils Imagen 3, a text-to-image model, on ImageFX for US users. While praised for high-quality images, critics point out strict content filtering. xAI's Grok-2 model on Elon Musk's X platform sparks controversy over AI ethics and content control. As AI image generation advances, tech firms must balance innovation with social responsibility for regulatory compliance and public trust.

Runway ML has launched the Gen-3 Alpha Turbo, 7x faster and 50% cheaper than its predecessor, making AI video creation more accessible. The Turbo model is now open to all users, including free trial users. Valenzuela, CEO of Runway, praised its speed. OpenAI's Sora model remains mysterious and inaccessible. Runway continues to lead the AI video generation market with faster and more cost-effective solutions.

Vision-language models (VLM) face challenges due to complex and fragmented evaluation methods. UniBench offers a standardized framework with 53 diverse benchmark tests to streamline VLM evaluation, revealing key insights for improving model performance. Its efficient workflow allows for rapid and comprehensive assessments.

Midjourney launches new unified web editor integrating key image editing tools, moving away from Discord roots. Offers message mirroring between Web Rooms and Discord. Aims to enhance usability for visual creators through a more accessible web platform. Recent improvements in version 6.1 include smoother images, faster processing, and enhanced text accuracy. Facing competition from models like Flux 1 and Imagen 3.

OpenAI closes SearchGPT prototype testing channel, limiting participation to 10,000 users worldwide. The move aims to collect feedback for algorithm optimization and avoid inappropriate recommendations. The cautious approach reflects the tech giant's focus on improving the generative search technology and providing valuable user experiences in the future.

Ant Group officially establishes "Digital Ant Force" in Beijing, focusing on AI to B market, offering tailor-made industry solutions to upgrade enterprise services efficiently. The platform provides intelligent customer service, operations, and technology services to help enterprises embrace AI effortlessly. Digital Ant Force's human-machine fusion AI solution has shown significant results in projects, leading the innovation in enterprise services. Exciting debut at the Bund Summit in September

Baidu Wenku launches AI app "Chengpian" with advanced features like multi-image composition and long article generation, revolutionizing content creation. It also plans to introduce AI-PPT, mind map, comic, and social media optimization capabilities, aiming to enhance user creativity.

In the realm of AI detection, Grammarly introduces Authorship to distinguish human-written from AI-generated content - aiming to mitigate inaccuracies seen in existing tools. With an initial launch on Google Docs, it sets sights on expanding. Targeting the education sector, Authorship plans to streamline the dialogue on AI's role in academia, offering clarity and transparency to students and instructors. Stay tuned to see if Grammarly's Authorship will outshine its competitors.

Google launches AI-generated search summaries in 6 new countries, displayed on the right side of search results. Added clickable links for direct website navigation. Users can save AI summaries for later viewing, with simplified versions available. Quality control measures in place to avoid issues encountered during earlier US launch.

OpenAI launched Advanced Voice Mode on ChatGPT for human-like voice conversations. Ethical and technical challenges arise. Users can choose from various voices like Juniper. This trend towards humanized human-machine interactions raises questions about AI ethics and future interaction models.