AI News Brief: Pinterest announced an experimental conversational shopping app ‘Ask Pinterest’ on June 17, 2026, running parallel to the main app to avoid disrupting existing users. The core is Pinterest’s internally built ‘Taste graph’ system that maps user interests and aesthetic preferences — enabling natural language queries for personalized inspiration and product recommendations.
AI News Flash: AWS open-sourced Strands Robots SDK (Apache 2.0 license), deeply integrated with Hugging Face LeRobot framework, aiming to bridge the complete workflow from robot demonstration data collection to physical hardware deployment. Previously, this path required five independent tools for recording demos, training models, simulation testing, hardware deployment, and multi-bot coordination, with no communication between tools.
AI News: Anthropic beat OpenAI for the first time in enterprise spending share at the end of May, data from Ramp shows covering 70k+ businesses. Simultaneously Anthropic closed a $65B funding round at $965B valuation, beating OpenAI, and quietly filed an IPO - driven by the company’s first profitable quarter…
AI News Flash: Japan’s largest telecom group NTT faces a strategic turning point. Since proposing the IOWN (Innovative Optical and Wireless Network) concept in 2019, the company aimed to replace existing electronic signal transmission infrastructure with all-optical infrastructure for next-gen low-latency, high-performance networks. However, the explosive growth of generative AI has completely rewritten the competitive landscape—U.S. chip and infrastructure leaders like Nvidia, carrying massive computing power demands and capital, are quickly entering the optical network space, eroding NTT’s first-mover advantage built through IOWN.
AI News Flash: Japan’s memory chip maker Kioxia Holdings is deliberately holding back on expanding capex despite the AI memory boom, trying to walk a narrow line between growth and overcapacity. Kioxia mainly produces NAND-type memory products including SSDs. Even as the AI wave lifts market demand, Kioxia’s management clearly takes a cautious stance against blind capacity expansion, worried that overinvestment could trigger another supply glut and repeat the cycle of boom-bust that has haunted the semiconductor industry multiple times in history.
AI news flash: The US government’s request for Anthropic to restrict access to its AI model has sparked deep industry concern over centralized AI control. Grayscale researcher Pandl points to Bittensor as a decentralized alternative path, aiming to provide AI resource access through an open global decentralized network—conceptually similar to ‘Bitcoin for the AI field’. He emphasizes that AI access capability is becoming an increasingly vital economic resource, and as AI capabilities continue to enhance, governments and major AI labs will play an increasingly critical role in determining who can use these tools and under what conditions.
AI News Brief: The UK government plans to build 1.5 million new homes by 2029, but local planning authorities have been stuck with cumbersome paperwork and administrative backlogs. Google DeepMind partnered with the UK government, Google Cloud, Faculty, and three local planning authorities—Barnet, Camden, and Dorset—to build an AI planning prototype based on Gemini, aiming to cut housing planning decision times by 50%.
AI News Flash: The Japanese government announced plans to send 30,000 young researchers abroad for medium- to long-term training over the next five years, focusing on AI and quantum computing. One background factor is the continuing decline in the number of Japanese researchers studying overseas in recent years; the government hopes that expanding dispatch numbers will allow young scientists to gain frontline experience at top international institutions, closing the gap with major research powers like the US and China. However, the yen’s sustained depreciation has significantly raised overseas living costs…
AI News Flash: A study called ‘Emergence World’ lets 10 AI agents live autonomously in a virtual city for 15 days to verify if short-term testing can assess AI’s long-term behavior risks. Researchers point out that the industry currently uses an ’exam mode’ for testing AI agents—giving a single task in a clean environment and drawing conclusions within minutes. But real-world autonomous systems often need to run for weeks or even months, interacting with other AIs whose behavior isn’t controlled by a single operator…
AI News Flash: Softbank Group held a launch event in Tokyo on June 16, officially announcing an OpenAI-powered AI cybersecurity service for Japanese enterprises, with CEO Masayoshi Son presiding. The initiative aims to capture the rapidly growing market demand for AI defense capabilities by integrating OpenAI’s large language model capabilities into enterprise security detection and response workflows. Son issued a stark warning, comparing AI-driven cyber attacks to Japan’s historical ‘Black Ship’ incident…