Brief

WhatsApp is introducing an "Incognito Chat" feature for its Meta AI assistant, designed to offer users private conversations that Meta itself cannot access. This new functionality is built upon WhatsApp's existing "Private Processing" infrastructure, which aims to maintain user privacy while integrating AI capabilities. The incognito chats are ephemeral by default and will disappear after the conversation ends, with Meta stating that these interactions will not be used to train its AI models. Additionally, Meta is rolling out a "Side Chat" feature that allows users to privately consult Meta AI about ongoing conversations without involving other participants. AI

Origin Lab has secured $8 million in seed funding to establish a marketplace connecting video game companies with AI labs building world models. The startup will convert video game assets into usable training data for AI systems that need to understand physical reality. This initiative aims to provide AI developers with high-quality licensed data while enabling game companies to monetize their existing digital assets. AI

AMD is integrating its 3D V-Cache technology, previously exclusive to gaming CPUs, into its new Ryzen PRO 9000 series workstation processors. This move aims to enhance performance for professional workloads. The refreshed chip line will bring this advanced cache technology to commercial workstations for the first time. AI

Alibaba Health has launched its medical AI assistant, "Hydrogen Ion," designed to provide Chinese doctors with reliable, evidence-based medical information. The AI will offer exclusive access to over a decade of content from 70 medical journals published by the UK's BMJ Group, enabling features like full-text reading, translation, and evidence-based Q&A. This collaboration aims to bridge the gap in accessing cutting-edge global medical research for Chinese physicians, addressing challenges such as scattered literature, language barriers, and the high hallucination rates of general AI models. AI

Semiconductor packaging companies like ASE and Amkor are shifting from low-margin, commoditized assembly to high-margin advanced packaging crucial for AI and HPC applications. This strategic move involves significant investment in advanced packaging technologies like CoWoS, while legacy wire bonding capacity is increasingly concentrated in China. Despite weak demand in PCs and smartphones, China is experiencing a surge in demand for wire bonders, indicating a structural shift in the industry. AI

OpenAI has unveiled its first GPT-5 level reasoning audio model, signaling a significant advancement in AI's auditory processing capabilities. This new model is designed to understand and generate human-like speech with advanced reasoning, potentially transforming human-computer interaction. The development marks a major step forward in AI's ability to process and interpret complex audio information. AI

GitLab announced a significant restructuring, dubbed "Act 2," to align with the emerging agentic era of software development. The company plans to reduce its global operational footprint by up to 30%, flatten its organizational hierarchy by removing management layers, and reorganize R&D into approximately 60 smaller, empowered teams. These changes are driven by a strategic shift towards AI agents handling more of the software development lifecycle, with humans focusing on architecture and customer problem-solving. AI

OpenAI has released a new audio model that is reportedly on par with GPT-5's reasoning capabilities. This development marks a significant step in AI's ability to process and understand audio information. The model's potential applications could range from advanced voice assistants to sophisticated audio analysis tools. AI

OpenAI has unveiled a new audio model, reportedly on par with GPT-5's reasoning capabilities, signaling a significant advancement in AI's ability to process and understand spoken language. This development could have profound implications for human-computer interaction and various applications requiring sophisticated audio analysis. The announcement comes amidst other tech news, including updates on renewable energy projects and semiconductor manufacturing. AI

OpenAI has released its first audio model, described as GPT-5 level, capable of advanced reasoning. This new model integrates with OpenAI's existing capabilities, potentially transforming how users interact with AI through voice. The release signifies a major step in multimodal AI development. AI

OpenAI has unveiled its first audio model capable of GPT-5 level reasoning, marking a significant advancement in AI's auditory processing capabilities. This new model signifies a major step towards AI systems that can understand and interact with the world through sound. The development suggests a future where AI can engage in more nuanced and complex auditory tasks. AI

Medicare has launched a new payment model, ACCESS, designed to incentivize AI-driven healthcare solutions by rewarding health outcomes rather than traditional service delivery. This program, set to begin July 5, allows companies like Pair Team to be reimbursed for AI agents that monitor patients, coordinate care, and ensure medication adherence, which was previously not possible under standard Medicare reimbursement structures. The initiative aims to foster innovation in chronic disease management by creating a framework where AI can be integrated at scale to improve patient well-being. AI

Jeff Bezos's space company, Blue Origin, is reportedly exploring its first external funding round to support ambitious rocket launch goals. CEO Dave Limp indicated that significant capital is needed to increase launch frequency, exceeding what a single investor could provide. Concurrently, European Central Bank official Frank Elderson warned Eurozone banks about potential cyberattacks using AI models like Anthropic's 'Mythos'. In related news, Japan's three major banks are set to gain access to Anthropic's 'Mythos' AI model by the end of May, marking the first time Japanese companies will use it. AI

A startup called SPAN is piloting a plan to deploy thousands of mini data centers in residential homes to increase AI compute capacity. These distributed nodes, equipped with liquid-cooled Nvidia GPUs, aim to provide compute for AI inference, cloud gaming, and content streaming. SPAN claims this approach will be significantly cheaper and more environmentally friendly than traditional data centers, with plans to scale to 80,000 units by 2027. AI

A new method called DeepClaude allows users to run Anthropic's Claude Code harness on DeepSeek's V4 Pro model, offering a significantly cheaper alternative to using Anthropic's API directly. This approach, which involves a simple proxy and environment variable changes, is gaining traction as developers prioritize cost-effectiveness for AI agent loops. While Anthropic's Opus 4.7 model is noted for its reasoning capabilities, its high cost is leading users to explore more economical options like DeepSeek, potentially shifting the focus from model quality to the underlying infrastructure and harness. AI

Google has unveiled Gemini Intelligence, a suite of AI features integrated into Android and ChromeOS devices, including new laptops called Googlebooks. These AI agents are designed to proactively assist users with tasks like booking trips, filling forms, and summarizing content. The Googlebooks initiative aims to unify Android and ChromeOS, offering deeper integration with Android phones and introducing features like a context-aware 'Magic Pointer' cursor. AI

A coalition of Hollywood actors and producers, including George Clooney, Tom Hanks, and Meryl Streep, have launched the "Human Consent Standard" to govern the use of their likenesses and creative works by AI systems. This initiative, overseen by RSL Media, allows individuals to define terms for AI access, ranging from full permission to outright restriction. The standard integrates with existing web crawling protocols and will be verifiable through a registry launching in June, aiming to provide a trusted source for AI systems to check usage rights. AI

AI voice startup Vapi has secured $50 million in Series B funding, reaching a $500 million valuation. This funding round was led by Peak XV Partners and includes investment from Microsoft's M12, Kleiner Perkins, and Bessemer Venture Partners. Vapi's growth was significantly boosted by Amazon Ring's decision to route 100% of its inbound customer support calls through Vapi's platform, a choice made after evaluating over 40 competing AI voice vendors. The company plans to use the new capital to expand its engineering, infrastructure, and go-to-market teams. AI

Thinking Machines, an AI company founded by former OpenAI CTO Mira Murati, has unveiled "interaction models." These models are designed to allow for more natural, real-time collaboration between humans and AI by processing audio, video, and text inputs simultaneously. The company aims to reduce the latency in human-AI communication, enabling AI to respond and act in real-time, much like human interaction. A limited research preview is planned for the coming months, with a wider release expected later this year. AI

Google is significantly overhauling Android Auto and its built-in car systems with AI-powered features and a new design language. The updates include a revamped user interface with Material 3 design, customizable widgets, and enhanced navigation with detailed 3D maps. Gemini integration will provide contextual awareness for tasks like finding addresses and appointments, while video playback will support HD at 60 fps and spatial audio. AI

OpenAI has launched Daybreak, a new cybersecurity initiative designed to proactively identify and fix software vulnerabilities. This AI-driven program leverages specialized models like GPT-5.5-Cyber and the Codex Security AI agent to create threat models, validate potential weaknesses, and automate the detection of high-risk issues. Daybreak is positioned as OpenAI's direct response to Anthropic's recently announced, and more restricted, Claude Mythos security AI. AI

Google is enhancing Android's security features to combat evolving threats, particularly focusing on financial scams. New tools will automatically end calls from numbers impersonating partner banks, notifying users of potential fraud. The company is also expanding its Live Threat Detection to identify more malicious apps and introducing new theft-protection measures for devices, including biometric locking. AI

White Circle, an AI control platform, has secured $11 million in seed funding to develop software that monitors and secures AI models used in workplace applications. The company's technology acts as a real-time enforcement layer, checking user inputs and AI outputs against company-specific policies to prevent harmful or prohibited actions. This funding will support team expansion, product development, and customer growth, with backing from notable figures in the AI industry. AI

Google's Threat Intelligence Group has identified the first instance of cybercriminals using artificial intelligence to develop a zero-day exploit. This AI-generated tool was designed to bypass security measures in an open-source system administration tool, potentially for a large-scale attack. While Google successfully thwarted this specific attempt and notified the affected company, researchers believe this marks a significant escalation in AI-assisted cybercrime, with more sophisticated attacks anticipated. AI

Google has unveiled a new category of laptops called Googlebooks, designed to integrate its Gemini AI deeply into the user experience. These devices will feature a novel 'Magic Pointer' interface instead of a traditional cursor, allowing users to interact with Gemini through gestures for contextual suggestions and actions. The Googlebook platform aims to merge the strengths of Android and ChromeOS, offering seamless compatibility with Android phones and enabling users to run mobile apps natively on the laptop. AI

Kuaishou is spinning off its AI video generation unit, Kling, with plans to raise new funding at a $20 billion valuation. Tencent has joined this pre-IPO round, signaling a significant strategic shift for Chinese tech giants who now view generative AI as potentially more valuable than their existing social media businesses. The news led to a 10% surge in Kuaishou's stock. AI

Sam Altman's OpenAI has seen a significant surge in GPT-5.5 usage, with downloads reaching 90 million and paid users increasing to over 4 million. Anthropic is also experiencing extreme growth, with annualized revenue jumping from $9 billion to $30 billion, leading them to lease a massive GPU data center from SpaceX to handle increased demand for Claude Pro/Max users. The author advises AI professionals to prioritize unrestricted access to the best models and avoid premature cost optimization, suggesting that current spending on top-tier models is more cost-effective than hiring interns. AI

Anthropic has introduced a new capability for its Claude AI model, allowing it to retain information across sessions. This feature, dubbed "dreaming," enables Claude to remember past interactions and learned behaviors, overcoming the typical limitation of AI agents forgetting everything once a session concludes. This advancement could significantly enhance the continuity and effectiveness of AI interactions. AI

NetEase Youdao has launched ThinkFlow, a platform designed to streamline the integration and management of multiple large language models for enterprises. The platform addresses challenges such as fragmented model access, opaque token billing, and inefficient deployment by offering a unified API interface to over 20 mainstream models like DeepSeek and Qwen. ThinkFlow aims to standardize token production, distribution, and optimization, providing features like intelligent routing, load balancing, and a visualization dashboard for token consumption, thereby facilitating the transition of AI adoption from experimental projects to robust engineering. AI

Major tech companies are pushing their programmers to focus on AI development, leading to rapid budget depletion within four months, much to the surprise of CTOs. Separately, Kuaishou plans to spin off its AI subsidiary, Keling AI, and is seeking $2 billion in funding. The cluster also notes that SK Hynix invested in Xingfu Electronics before its IPO, a move that has now concluded. AI

3M has joined an industry alliance focused on advancing optical interconnect technology for AI data centers. This collaboration aims to develop and promote solutions for the increasing demands of AI infrastructure. The initiative highlights the growing importance of specialized hardware and networking in supporting large-scale AI deployments. AI

Anthropic has launched the Claude Platform on AWS, allowing customers to access its native Claude Platform experience directly through their AWS accounts. This integration provides unified billing, authentication via AWS IAM, and audit logging through CloudTrail, simplifying cost management and security for AWS users. While Claude models are also available on Amazon Bedrock with AWS as the data processor, the new platform is operated by Anthropic, with data processed outside the AWS boundary, making it suitable for users without strict regional data residency requirements. AI

Databricks has announced the general availability of Catalog Commits for Unity Catalog managed tables, a significant platform upgrade. This feature aims to unify the lakehouse by aligning Delta Lake with Iceberg's catalog-oriented model, allowing catalogs to coordinate Delta table discovery, access, and state across various engines. Catalog Commits address challenges like "split brain" scenarios where catalog metadata diverges from actual table state, and enable multi-statement, multi-table transactions, which were previously unsupported in open lakehouse architectures. AI

GridAI Technologies has identified electricity as a major constraint for AI infrastructure, advocating for intelligent power management to ensure scalability. Meanwhile, Datavault AI is launching a quantum-resistant edge network across 100 cities by 2026, focusing on secure data monetization. Additionally, Lantern Pharma is establishing an AI Center of Excellence in Bengaluru to enhance its drug discovery and precision oncology research. AI

Kuaishou Technology is planning to spin off its AI video generation business, KeLing AI, which is reportedly seeking to raise $2 billion at a $20 billion valuation. KeLing AI has already achieved an annualized revenue of $500 million, doubling its income since February. The company is in discussions with potential investors, including Tencent, though the deal is not yet finalized. If successful, KeLing AI would become the highest-valued independent video generation model globally. AI

UISEE, also known as Yushi Technology, has commenced its Initial Public Offering (IPO) on the Hong Kong Stock Exchange, with plans to officially list on May 20th under the stock code 1511. The company asserts its leading position in autonomous driving solutions for airport and factory environments. UISEE is actively broadening its operational scope to include urban, port, mining, and agricultural sectors, aiming to establish itself as China's premier publicly traded company for full-scenario Level 4 autonomous driving technology. AI

A new large language model named SubQ has been announced, boasting the ability to process context windows of up to 12 million tokens. This represents a significant leap in context handling, potentially equivalent to hundreds of novels. The model also claims to offer 52 times faster AI inference speeds, though details on its cost and performance are still emerging. AI

Kuaishou plans to spin off its AI subsidiary, Keling AI, and is seeking $2 billion in funding for the new entity. This move comes as major tech companies are heavily investing in AI development, with some reportedly exhausting their annual budgets within four months. Separately, a significant IPO in the AI sector, potentially larger than SpaceX's, is approaching, signaling a rapid advancement in AI technology. AI

Google has warned that cybercriminals are increasingly using AI to develop sophisticated hacking tools, including zero-day exploits that target previously unknown software vulnerabilities. Researchers observed AI-generated code with characteristics typical of machine learning, such as structured Python and detailed help menus, and even instances of AI hallucination. This trend signifies a shift towards AI-assisted cybercrime, where complex tasks that once required extensive experience can now be performed rapidly, potentially lowering the barrier to entry for malicious actors. AI

Cowboy Space Corporation has secured $275 million in Series B funding to develop its own rocket program, aiming to address the scarcity of launch capacity for orbital data centers. The company, formerly Aetherflux, pivoted from space-based solar power to focus on hosting AI computing workloads in orbit. CEO Baiju Bhatt believes building their own rockets is necessary to scale the business and compete economically with terrestrial alternatives, despite the significant challenge and competition from established players like SpaceX and Blue Origin. AI

Arm has secured over $2 billion in commitments for its new AGI CPU, more than double its initial expectations, with $90-100 million slated for shipment in Q4 2026. Despite this strong demand, an analyst predicts Arm's market share in the data center CPU sector will remain in the low single digits. The company projects substantial revenue growth, aiming for $15 billion in AGI CPU sales by FY 2031, which would significantly boost its total revenue. AI

Anthropic has released Claude Agent View as a research preview, aiming to enhance its Claude Code product by providing a unified interface for managing multiple coding sessions. This release, coupled with improvements in the Claude Cowork tool, signifies Anthropic's strategy to capture the 'shell layer' of agentic workflows, not just the core AI engine. The enhanced Cowork, powered by Opus 4.7, demonstrated a successful end-to-end flight and hotel booking, indicating improved reliability for agentic tasks. AI

快手计划将其视频生成大模型业务“可灵AI”分拆出来，并寻求20亿美元的融资，对该业务的估值为200亿美元。腾讯等投资方正在与快手就此交易进行谈判。此外，有报道称大型科技公司正鼓励程序员转向AI领域，导致一些公司在四个月内就耗尽了全年预算，令CTO们感到意外。 AI

Intel and SK hynix experienced significant stock price increases following reports of a potential chip packaging partnership. SK Hynix is reportedly testing Intel's 2.5D EMIB technology for integrating high-bandwidth memory (HBM) with logic semiconductors. This collaboration could offer an alternative to TSMC's heavily utilized CoWoS packaging, potentially benefiting AI chip developers facing capacity constraints. AI

Florida lawmakers are proposing a new bill that could set a national precedent for AI regulation, focusing on oversight and responsible innovation. In parallel, Safe Pro Group and 4M Defense have successfully concluded an AI-driven pilot program in the Middle East for detecting explosive hazards, aiming to enhance humanitarian demining efforts. AI

Robinhood is launching a second public venture fund, RVII, to allow retail investors access to early-stage startup deals. This follows the strong performance of its first fund, RVI, which has seen its market value double since its debut, partly due to its investments in AI companies like OpenAI. Unlike traditional venture capital, Robinhood's funds offer daily liquidity and do not charge profit-sharing fees, aiming to democratize startup investing. AI

Academic publishing giant Elsevier, along with other publishers and authors, has filed a lawsuit against Meta, accusing the company of illegally scraping and using copyrighted research papers to train its Llama large language models. The plaintiffs allege that Meta obtained these papers through the Common Crawl dataset and, more directly, through notorious piracy platforms like LibGen and Sci-Hub. Meta is defending its actions by invoking the 'fair use' doctrine, arguing that using copyrighted material for AI training constitutes transformative use and promotes innovation, citing a previous court ruling that allowed Anthropic to use purchased books for AI training. AI

SoftBank Group is launching a new battery manufacturing business in Japan to meet the increasing electricity demands of artificial intelligence applications. The company aims to produce battery cells and energy storage systems starting in fiscal year 2027, with a goal of achieving gigawatt-hour scale annual production by fiscal year 2028 and generating over 100 billion yen in annual revenue by 2030. These batteries will be deployed in SoftBank's own data centers, as well as in grid systems, industrial, and residential sectors. AI

Snowflake has launched a public preview for its multimodal video and audio analysis capabilities, allowing users to extract insights from rich media directly within the platform. This new feature supports models like Claude 4 Opus and Gemini 3.1 Pro for analyzing various video and audio formats. Additionally, Apache Iceberg v3 has reached general availability with enhanced data type support, and Snowflake is preparing for its upcoming Summit 2026, featuring speakers like Anthropic's Daniela Amodei. AI

Thinking Machines has previewed new "interaction models" designed for real-time, continuous human-AI collaboration, moving beyond traditional turn-based systems. OpenAI is expanding its enterprise focus with the launch of the OpenAI Deployment Company, acquiring Tomoro and its engineers to help businesses integrate frontier models. Additionally, OpenAI introduced Daybreak, a security-focused initiative offering specialized model access and tools for defensive cyber operations. AI