Welcome to the August 11, 2025 edition of ACM TechNews, providing timely information for computer professionals three times a week.

HEADLINES AT A GLANCE

• Japan Launches Fully Domestically Produced Quantum Computer
• Microsoft Exchange Server Vulnerability Enables Attackers to Gain Admin Privileges
• A Single Poisoned Document Could Leak 'Secret' Data via ChatGPT
• Taiwan Cultivates Young Overseas Chip Talent With Summer Camps, University Courses
• Research by U.S. Engineers Improves Strength Prediction in 3D Printing
• China Plots Pathway to Tech Supremacy through Brain-Computer Interfaces
• Researcher Deploys Fuzzer to Test Autonomous Vehicle Safety
• The Militarization of Silicon Valley
• Developers Are Frustrated with AI Coding Tools That Deliver Nearly Right Solutions
• World’s First Humanoid Robot 4S Store Opens in Beijing
• Memories of Deadly Winter Blackouts Haunt the Texas Datacenter Boom
• The Bright Side: Underwater Robot Live Stream Mesmerizes Argentines

Japan Launches Fully Domestically Produced Quantum Computer

A superconducting quantum computer launched by Japan's University of Osaka Center for Quantum Information and Quantum Biology (QIQB) was designed and built with domestically produced software and components. QIQB collaborated with RIKEN, ULVAC, QuEL, QunaSys, Fujitsu, and other companies to replace previously imported parts, such as the dilution refrigerator, with domestically manufactured parts. The software uses the open source OQTOPUS framework. The quantum computer will be available in a dedicated exhibit at Expo 2025 in Osaka.
[ » Read full article ]

Microsoft Exchange Server Vulnerability Enables Attackers to Gain Admin Privileges

A critical vulnerability (CVE-2025-53786) in Microsoft Exchange Server hybrid deployments allows attackers with on-premises admin access to escalate privileges to Exchange Online without leaving clear audit traces. Demonstrated at Black Hat 2025, the flaw stems from shared service principals in hybrid authentication. Microsoft began mitigation in April 2025 by introducing dedicated hybrid applications, later formalizing the issue in this CVE.
[ » Read full article ]

A Single Poisoned Document Could Leak 'Secret' Data via ChatGPT

A vulnerability in OpenAI's ChatGPT Connectors allows sensitive information to be extracted from Google Drive via an indirect prompt injection attack called AgentFlayer, revealed researchers Michael Bargury and Tamir Ishay Sharbat of Zenity during a recent session at Black Hat USA 2025. The exploit involves hiding a malicious prompt in a shared document, unseen by humans but executed by the AI, causing ChatGPT to leak data.
[ » Read full article ]

Taiwan Cultivates Young Overseas Chip Talent With Summer Camps, University Courses

Taiwan is addressing its semiconductor talent shortage by launching bilingual camps and new university programs to attract young international students. U.S.-based Synopsys is hosting summer camps in both Mandarin and English to spark early interest in chip design, while institutions such as National Taiwan University and NYCU have introduced global education tracks and outreach initiatives.
[ » Read full article ]

Research by U.S. Engineers Improves Strength Prediction in 3D Printing

A method for more accurate strength prediction in 3D-printed parts using gyroid infill patterns has been developed by researchers at the University of Maine. By combining computer simulations with physical stress testing, the team validated finite element analysis models that assess performance under compression and shear. This approach enables more reliable design decisions and supports material efficiency in applications where strength-to-weight ratio is critical.
[ » Read full article ]

China Plots Pathway to Tech Supremacy through Brain-Computer Interfaces

China's 2027 technology roadmap emphasizes brain-computer interfaces (BCI) as a key area in its technology competition with the U.S. The Ministry of Industry and Information Technology and six other Chinese agencies have released a policy blueprint for BCI, including targets for breakthroughs in chips, electrodes, and integrated products. Notable research includes a minimally invasive vascular intervention electrode to improve grasping in a stroke patient, an implantable processor aiding quadriplegic patients, and an ultra-flexible, coin-sized electrode with fast, stable neural decoding for human-machine interactions.
[ » Read full article ]

Researcher Deploys Fuzzer to Test Autonomous Vehicle Safety

Research using a fuzzing framework to test autonomous vehicle teleoperation systems against unsafe or malicious commands was detailed during a session by Zoox product security engineer Zhisheng Hu at Black Hat USA 2025. The technique generated edge cases that could cause collisions, revealing vulnerabilities that might appear benign until specific conditions arise. The findings highlight fuzzing as a key method for scaling autonomous vehicle safety and security in real-world conditions.
[ » Read full article ]

The Militarization of Silicon Valley

Big Tech executives including Andrew Bosworth (Meta CTO), Shyam Sankar (Palantir CTO), Kevin Weil (OpenAI CPO), and Bob McGrew (advisor at Thinking Machines Lab and former OpenAI chief research officer) were sworn in as lieutenant colonels as part of their participation in Detachment 201, a technical innovation unit created by the U.S. Army. The unit will advise the Army on new combat technologies, illustrating a growing trend in Silicon Valley in which companies and venture capitalists increasingly engage with military technology and remove corporate policies that prevent AI use in weapons.
[ » Read full article *May Require Paid Registration ]

Developers Are Frustrated with AI Coding Tools That Deliver Nearly Right Solutions

A survey of 49,009 developers across 160 countries found widespread use of AI coding tools, but limited trust in them. Although 78.5% of respondents reported using AI tools at least occasionally, only 3.1% expressed some trust in their output. Developers cited frustration with tools producing “almost right” code and difficulties debugging. Complex tasks remain a major weakness, and many rely on humans when accuracy or understanding is critical.
[ » Read full article ]

World’s First Humanoid Robot 4S Store Opens in Beijing

The world’s first 4S store for humanoid robots, Robot Mall, has opened in Beijing’s Yizhuang alongside a robot-themed restaurant, coinciding with the 2025 World Robot Conference. The 4,000-square-meter venue showcases more than 50 brands across seven categories, with services from selection to maintenance. Exhibits span medical, industrial, bionic, and service robots, with interactive zones and business matchmaking. Yizhuang, home to 300 robotics firms and half of Beijing’s robotics output, is emerging as a leading innovation hub.
[ » Read full article ]

Memories of Deadly Winter Blackouts Haunt the Texas Datacenter Boom

Texas has enacted a first-of-its-kind law requiring large datacenters to install emergency shutoff switches, enabling the state grid operator to cut their power during crises. The measure follows deadly 2021 blackouts and comes amid surging electricity demand from AI-driven facilities, including the $500 billion Stargate project backed by President Trump.
[ » Read full article *May Require Paid Registration ]

The Bright Side: Underwater Robot Live Stream Mesmerizes Argentines

An Argentinean and American 20-day expedition deployed an advanced underwater SuBastian robot capable of descending nearly 4,500m into the Mar del Plata canyon off Argentina. Most of the researchers are from Argentina's CONICET, and the expedition is streaming live high-definition video of the deep ocean floor to a broad audience via YouTube and TV. Operated remotely from the Schmidt Ocean Institute's research vessel, the robot is equipped with robotic arms and sampling tools to collect unique biological specimens and capture striking imagery of previously undocumented marine environments.
[ » Read full article ]