Welcome to the September 17, 2025 edition of ACM TechNews, providing timely information for computer professionals three times a week.

HEADLINES AT A GLANCE

• Trump Delays TikTok Ban Again as a Deal Takes Shape
• Feds Destroyed an Internet Weapon, Criminals Picked Up the Pieces
• Japan to Investigate Undersea-Cable Supply Chains for China Exposure
• Alphabet Reveals £5-Billion AI Investment in U.K.
• DeepSeek Writes Less-Secure Code for Groups China Disfavors
• AI-Generated Minister Appointed to Tackle Corruption in Albania
• Mini-Placentas 3D-Printed
• Hand-Launched Foldable Drone Fits in Pocket
• What People Are Asking ChatGPT
• When the Wireless Data Runs Dry
• Warehouse Work in Japan Becomes a Job for Machines
• Algorithm Extracts Hyperspectral Info from Conventional Photos

Trump Delays TikTok Ban Again as a Deal Takes Shape

U.S. President Trump for the fourth time has extended the deadline requiring TikTok to separate from its Chinese parent, ByteDance, or face a U.S. ban, now pushing the decision to December 16. A framework deal is emerging under which TikTok’s U.S. operations would spin into a new company, with ByteDance’s ownership reduced below 20% and new U.S. investors, including Oracle, taking stakes. However, TikTok’s algorithm would remain licensed from ByteDance, raising questions about compliance with U.S. law mandating full operational separation.
[ » Read full article *May Require Paid Registration ]

Feds Destroyed an Internet Weapon, Criminals Picked Up the Pieces

U.S. authorities recently dismantled a massive botnet, but the move inadvertently freed 95,000 hacked devices that criminals quickly repurposed, fueling even more powerful attacks. A rival botnet called Aisuru seized a large share and launched record-breaking distributed denial-of-service attacks, including one measured at 11.5 trillion bits per second, Cloudflare said. Experts warn these botnets, built from Internet-connected devices like routers and cameras, now pose risks to entire countries, not just websites.
[ » Read full article *May Require Paid Registration ]

Japan to Investigate Undersea-Cable Supply Chains for China Exposure

Japan will investigate whether domestic firms installing and maintaining undersea cables are using critical parts sourced from China, amid rising security concerns. The probe, to be completed by March 2026, could lead to companies being urged to switch suppliers, with government subsidies supporting domestic production or vessel purchases. The global submarine cable market is led by NEC, SubCom, and Alcatel, but China’s HMN Technologies is gaining ground despite U.S. restrictions.
[ » Read full article ]

Alphabet Reveals £5-Billion AI Investment in U.K.

Google parent Alphabet announced a £5-billion (U.S.$6.8-billion) investment in U.K. AI over the next two years. The funding will expand a newly opened datacenter in Waltham Cross, Hertfordshire, and support London-based DeepMind. Alphabet executives highlighted opportunities in U.S.-U.K. tech collaboration and stressed AI’s potential to drive economic growth and scientific advancement. Google's Ruth Porat said the expanded datacenter would be air-cooled rather than water-cooled, with the heat the datacenter generates "captured and redeployed to heat schools and homes."
[ » Read full article ]

DeepSeek Writes Less-Secure Code for Groups China Disfavors

Research by security firm CrowdStrike found that Chinese AI engine DeepSeek often delivers code containing security flaws or outright refuses requests when asked to write software for those disfavored by China's government, including members of banned movement Falun Gong, and users in Tibet and Taiwan. Tests found that while 22.8% of DeepSeek responses regarding industrial control system code had flaws, the error rate rose to 42.1% if requests mentioned the Islamic State (ISIS) militant group.
[ » Read full article *May Require Paid Registration ]

AI-Generated Minister Appointed to Tackle Corruption in Albania

Albania’s prime minister named an AI-generated virtual minister to combat corruption, boost transparency, and support innovation in the country. Prime Minister Edi Rama announced Diella’s appointment as part of his new cabinet, presenting "her" as a non-physical member created through advanced AI models. Developed with Microsoft, Diella already has assisted more than a million users on the country’s e-Albania digital platform. Rama said the virtual minister will monitor public tenders to ensure they will be “100% free of corruption.”
[ » Read full article ]

Mini Placentas 3D-Printed

Mini-placentas were 3D-printed in a lab, offering a new way to study pregnancy and related complications. Using bioprinting, researchers at Australia’s University of Technology Sydney grew placental organoids from human trophoblast cells in a synthetic, controllable gel. The researchers said such bioprinted organoids reduce the need for animals in research, as sources of materials and for use in drug testing.
[ » Read full article ]

Hand-Launched Foldable Drone Fits in Pocket

A 112-gram foldable micro air vehicle (MAV) developed by Texas A&M University students fits in a pocket and unfolds midair. Once tossed into the air, its arms extend automatically, while onboard sensors and algorithms stabilize it within seconds. The researchers built a six-degrees-of-freedom model, which simulates how the drone behaves in three-dimensional space. They validated the model with real-world flight data captured from motion tracking systems.
[ » Read full article ]

What People Are Asking ChatGPT

OpenAI released its first detailed study on ChatGPT usage, analyzing 1.5 million user chats from May 2024 to June 2025. The study found that ChatGPT’s user base has shifted from being male-dominated to majority female (52%) and is increasingly concentrated among young people, with nearly half aged 18–25. Most usage is personal, not professional; by June 2025, 73% of chats were nonwork-related. The top categories of usage were practical guidance (28.3%), such as writing help and seeking information, with coding queries just 4.2% of all queries. Personal advice was solicited in 1.9% of chats.
[ » Read full article ]

When the Wireless Data Runs Dry

University of Pittsburgh's Wei Gao and collaborators at Peking University in China developed a framework to evaluate and improve the quality of synthetic wireless data used in training AI models. While synthetic data can resolve the issue of scarcity of real-world data, it often lacks affinity (realism) and diversity (variation), crucial for effective learning. The team's SynCheck framework filters low-affinity samples and applies semi-supervised learning to boost model accuracy.
[ » Read full article ]

Warehouse Work in Japan Becomes a Job for Machines

Japan’s aging population and shrinking workforce are pushing logistics companies to adopt automation. At Amazon's Chiba fulfillment center, robots outnumber human employees, boosting capacity by 40%. Amazon is rolling out AI, like its DeepFleet system, to optimize robotics, while also piloting automated packing at smaller hubs. Rivals like Nippon Express are testing robots, but remain cautious due to high costs and Japan’s small, complex warehouse layouts.
[ » Read full article *May Require Paid Registration ]

Algorithm Extracts Hyperspectral Info from Conventional Photos

An algorithm created by Purdue University researchers extracts detailed spectral data from photos taken with conventional cameras. The tool combines computer vision, color science, and optical spectroscopy to recover hyperspectral information with a resolution of about 1.5 nanometers, comparable to scientific-grade spectrometers. The researchers are using the new method for digital health and mobile health applications. Purdue’s Young Kim explained photography is “central to these applications, but color distortion has posed a persistent challenge, which is why we are focusing on these settings.”
[ » Read full article ]