Today's Headlines
- Japan launches FRONTia, a national AI foundation-model project led by METI and NEDO — roughly 27,500 of NVIDIA's new Rubin GPUs, with a fiscal 2026 budget of 387.3 billion yen
- The European Commission formally adopts decisions forcing Google to share search data and open Android's AI features — data sharing from January 2027, Android changes by July 2027
- Google renames NotebookLM to Gemini Notebook — a secure cloud compute environment brings code execution and data analysis into every notebook
It is a day when states moved the board in AI.
Japan turned the development of homegrown foundation models into a full national project, laying out a single plan that runs from GPU procurement to a 2040 target. The EU, meanwhile, formally ordered Google to open up the two main gateways to AI — search and the smartphone. Industrial strategy on one side and market regulation on the other moved on the same day.
Today's Top Three
Japan Launches National AI Project FRONTia — Roughly 27,500 of NVIDIA's New Rubin GPUs
FRONTia, a national AI foundation-model development project led by Japan's Ministry of Economy, Trade and Industry (METI) and the public R&D agency NEDO, has launched.
FRONTia stands for "Foundation for Real-world Omni-Native Trustworthy Intelligence & Alignment." The project aims to build multimodal foundation models for physical AI — AI robots and machines that act in the real world. METI announced the start of the program on June 30, selecting two implementers: Noetra and AIST, Japan's National Institute of Advanced Industrial Science and Technology.
The structure is two-track. Noetra — a new company backed by 44 investors including SoftBank, Sony Group, NEC, and Honda — handles the development track, while AIST runs the exploratory research track. Daisuke Okanohara of Preferred Networks, one of Japan's best-known AI startups, joins as co-head of research and development.
The compute is now settled as well. NVIDIA announced on July 16, together with the Japanese government and industry partners, what it calls the world's first national AI infrastructure: Noetra will build an AI factory configured with roughly 27,500 of NVIDIA's new-generation Rubin GPUs and more than 13,750 Vera CPUs.
On money and timeline, the fiscal 2026 budget is 387.3 billion yen, the project runs from fiscal 2026 through fiscal 2030, and stage-gate reviews will be held every year from fiscal 2027. Beyond that sits a long-range goal: 10 million AI robots deployed in Japan by 2040.
- Japan launches national AI foundation-model project FRONTia with about 27,500 new NVIDIA GPUs (PC Watch, Japanese)
- METI launches the multimodal foundation-model development program for AI robots and physical AI (METI, Japanese)
- Japan Government, Industrial Leaders and NVIDIA Launch the World's First National AI Infrastructure (NVIDIA, Japanese)
EU Formally Orders Google to Share Search Data and Open Up AI on Android — Phased In Through 2027
The European Commission on July 16 formally adopted two "specification decisions" under the Digital Markets Act (DMA), requiring Google to share search data and to open Android's AI features to rivals.
The search decision rests on Article 6(11) of the DMA. Google must provide anonymized search ranking, query, click, and view data to competing search engine providers on fair, reasonable, and non-discriminatory terms. Providers of AI chatbots with search functions are also eligible for the shared data, and sharing must begin by January 2027.
The Android decision rests on Article 6(7). Google must give third-party AI assistants effective interoperability with the Android features its own Gemini enjoys — including activation by voice wake words and the ability to execute tasks across apps. "Third-party AI assistants are therefore limited in how they can offer their innovative services, making them less attractive to 60% of EU users who have an Android device," the Commission said. The deadline for the Android changes is July 2027.
Procedurally, the Commission opened the proceedings on January 27, set out its proposed measures on April 27, and has now adopted the final decisions within the statutory deadline. Specification decisions are legally binding, and Google, designated a "gatekeeper" under the DMA, must comply.
Google is pushing back hard. "Today's decisions risk undermining vital privacy and security guardrails for millions of Europeans," said Kent Walker, the company's president of global affairs. The Commission, for its part, is reported to remain open to amending the decisions on how personally identifiable data is handled.
- It's official: EU will force Google to share search data and open up AI on Android (Ars Technica)
- DMA.100220 — Alphabet - OS - Google Android - Art. 6(7) (European Commission case page)
- DMA.100209 — Consultation on the proposed measures for Google Search data sharing (European Commission)
NotebookLM Becomes Gemini Notebook — Cloud Compute Brings Code Execution Into the Notebook
Google is renaming its AI research tool NotebookLM to Gemini Notebook and adding a secure cloud compute environment to every notebook.
With the new environment, Gemini Notebook can write and execute code natively, carrying out complex data analysis grounded in the sources loaded into the notebook. It extends the product from a tool you talk to about your documents into one that does the computation itself.
The rollout starts with Google AI Ultra users and Workspace business customers on eligible licenses, with all Pro users on the web following in the coming weeks. Notebooks can now be created and accessed directly inside the Gemini app, with full syncing against the standalone product, and integration into AI Mode in Google Search is planned.
NotebookLM traces back to Project Tailwind, a research tool unveiled at Google I/O 2023, and now serves more than 30 million users and over 600,000 organizations. The renaming folds an established product into the Gemini brand, and it continues as a standalone research tool.
Other Moves
Models & Products
- China's Moonshot AI quietly released Kimi K3, a new flagship with 2.8 trillion parameters and support for up to 1 million tokens of context, live on kimi.com without prior announcement. Model weights and a technical report are expected within days. - Moonshot AI's 2.8-trillion-parameter Kimi K3 debuts with 1M-token context (PC Watch, Japanese)
- LM Studio unveiled LM Studio Bionic, a standalone AI agent built for open-source models. It handles coding assistance, document processing, and voice input locally, under a zero-data-retention policy. - LM Studio Bionic: the AI agent for open models (LM Studio)
- Google added two updates to Google Vids: Gemini Omni integration for generating and editing video through natural language, and personal avatars that let users appear in videos built from a selfie and a voice recording. Avatars are limited to users 18 and over in eligible regions, with SynthID watermarking applied automatically. - Create, edit and star in videos with two Google Vids updates (Google)
- Roblox announced Build, an AI feature in its mobile app that generates playable games from a text prompt alone. Public alpha testing starts July 28 in New Zealand. - Roblox launches an AI-powered game-creation feature in its mobile app (TechCrunch)
Research
- Google DeepMind and Isomorphic Labs published a "bioresilience" strategy aimed at preventing misuse of their AI models for biological weapons while helping governments and scientists use them for defense. The companies say they have built more than 15 partnerships with government and biosecurity organizations over the past 12 months. - Our approach to bioresilience (Google DeepMind)
Business
- Cars24, one of India's largest used-car marketplaces, said its OpenAI-powered voice and chat agents now handle more than one million conversation minutes a month. Support resolution rates rose 50%, and the ChatGPT Enterprise and Codex deployment across roughly 600 employees is seeing 85-90% daily active usage. - How Cars24 scales conversations and builds faster with OpenAI (OpenAI)
Policy & Other
- OpenAI published its approach to teen use of ChatGPT, pairing stronger learning features with age-appropriate safeguards. The company says nearly nine in ten teen users engage with the product weekly for learning and skill-building, and points to restricted experiences for users estimated to be under 18 and expanded parental controls. - Why teens deserve access to safe AI (OpenAI)
- Meta launched a feature that alerts parents using its supervision tools when a teen's conversation with Meta AI shows signs of self-harm or suicide risk, now live in the US, UK, Australia, and Canada. A global rollout is planned by the end of 2026. - Alerting Parents if Teens Show Signs of Distress in Conversations With Meta AI (Meta)
- Hyundai autoworkers in South Korea began a three-day rolling strike over wage demands and fears about the planned introduction of Atlas humanoid robots. It is reported as the first car-factory strike in which humanoid robots are a central issue. - Fear of humanoid robots spurs human workers to strike at Hyundai auto factory (Ars Technica)
- Linux creator Linus Torvalds pushed back against anti-AI sentiment in the kernel community, saying Linux is not an anti-AI project and telling critics to fork the kernel or walk away. The remarks make clear that one of open source's flagship projects is accepting AI assistance rather than resisting it. - Linus Torvalds to critics of AI coding in Linux: "Fork it. Or just walk away." (Ars Technica)
Source: Selected by the editorial team from the AI news inbox (30 items collected on July 17, 2026 — 9 primary, 21 secondary).