Banner
Welcome to the October 10, 2025 edition of ACM TechNews, providing timely information for computer professionals three times a week.

Please note: In observance of the U.S. Indigenous Peoples’ Day/Columbus Day holiday, TechNews will not be published on Monday, Oct. 13. Publication will resume on Wednesday, Oct. 15.

Mette Frederiksen Denmark plans to ban social media use by children under 15. Prime Minister Mette Frederiksen told Parliament that social media harms young people’s mental health, causing anxiety, depression, and concentration problems. Under the proposal, parental permission would be required for children to allow access to social media from age 13. Said Frederiksen, "Never before have so many children and young people suffered from anxiety and depression." The ban could take effect next year.
[ » Read full article ]
The Guardian (U.K.); Miranda Bryant (October 7, 2025)

Expanding a commercial robotic prosthetic leg’s benefits with advanced control University of Michigan researchers showed that the addition of a control algorithm to passive prosthetic legs can help both higher- and lower-mobility amputees with tasks like walking long distances or rising from a chair. The researchers developed an algorithm that was added to the Össur Power Knee. Unlike Össur's controller, which recognizes certain features of the user's motion that indicate their intent, the researchers' controller is based on mathematical models of human movement and continually adjusts to the user's motion.
[ » Read full article ]
Michigan Engineering News; Kate McAlpine (October 6, 2025)

City College of San Francisco is one of 116 California Community Colleges rolling out Nectir's AI learning assistants Through a partnership with AI firm Nectir, California Community Colleges will offer AI tutors to students and staff across its 116 campuses at no cost. Nectir's AI learning assistant provides 24/7 tutors that offer conversational and personalized feedback and guidance via built-in chatbots that provide coaching on financial aid, career prep, and more.
[ » Read full article ]
Axios; Shawna Chen (October 6, 2025)
A report by security firm LayerX found that of the 45% of enterprise employees using generative AI tools, 77% copy and paste data into ChatGPT queries. According to the study, 22% of those copying and pasting are doing so with personally identifiable information (PII) and payment card industry (PCI) data. The report said that enterprises “have little to no visibility into what data is being shared, creating a massive blind spot for data leakage and compliance risks.”
[ » Read full article ]
The Register (U.K.); Thomas Claburn (October 7, 2025)

Proposed FedLLMGuard Architecture A framework developed by researchers at the U.K.'s University of Portsmouth combines federated learning and large language models to identify flaws in 5G networks and provide real-time data protection. In tests against large-scale cyberattacks, data poisoning attacks, stealth attacks, and others, the FedLLMGuard framework was found to be 98.64% accurate in detecting threats quickly.
[ » Read full article ]
University of Portsmouth (U.K.) (October 8, 2025)

a screen displays guidelines for using artificial intelligence at Valencia High School in Santa Clarita, Calif., A new report from the Center for Democracy and Technology warns that the rapid rise of AI use in K-12 classrooms is creating serious downsides for students. About 85% of teachers and 86% of students used AI during the 2024–25 school year, according to the report. Nearly half of students, meanwhile, reported feeling less connected to teachers due to AI use, as well as reporting a decrease in peer-to-peer connections.
[ » Read full article ]
Education Week; Jennifer Vilcarino; Lauraine Langreo (October 8, 2025)

The test rig HydroHaptics, a responsive technology developed by researchers at the U.K.'s University of Bath, lets users perform certain actions and receive tactile feedback by tapping, twisting, or pinching a soft object. The researchers integrated HydroHaptics into a cushion to control smart home devices, a 3D joystick to provide gamers with haptic feedback, a backpack to deliver smartphone notifications or enable navigation, and a computer mouse to sculpt digital objects on a screen.
[ » Read full article ]
University of Bath (U.K.) (October 8, 2025)

The federal safety agency said its investigation of safety issues affects nearly 2.9 million vehicles Federal regulators have opened an investigation into Tesla’s “Full Self-Driving” (FSD) system after reports that it caused vehicles to run red lights and veer into oncoming lanes. The National Highway Traffic Safety Administration said the probe covers nearly 2.9 million Teslas following 58 reported incidents, including 14 crashes or fires and 23 injuries.
[ » Read full article *May Require Paid Registration ]
The Washington Post; Faiz Siddiqui (October 9, 2025)
Google's new AI Vulnerability Reward Program will reward security researchers who identify and report vulnerabilities in its AI systems. The bug bounty program covers high-impact issues in Google Search; Gemini Apps; Google Workspace core applications (such as Gmail, Drive, Meet, and Calendar); Google AI products, like AI Studio and Jules; Google Workspace non-core apps; and other AI integrations in Google products. The rewards include $5,000 for identifying phishing enablement and model theft issues, $15,000 for sensitive data exfiltration bugs, and up to $30,000 for individual quality reports with novelty bonus multipliers.
[ » Read full article ]
BleepingComputer; Sergiu Gatlan (October 7, 2025)
Researchers at North Carolina State University identified a hardware timing vulnerability in AI accelerators that can leak training data and other private information. The GATEBLEED vulnerability exploits power-gating behaviors in on-chip accelerators like Intel AMX, producing observable timing differences when models encounter data on which they were trained. It can be executed without special permissions, bypasses many existing defenses, and works against popular ML libraries and architectures. Mitigation requires hardware redesigns or costly OS/microcode fixes.
[ » Read full article ]
NC State News; Matt Shipman (October 8, 2025)

A machine-tending ‘cobot’ manipulates materials in and out of a computer numerical control mill The U.S. manufacturing revival is being driven by collaborative robots (cobots), which are smaller, cheaper, and easier to program than traditional industrial robots. Companies like Raymath in Ohio and Caltech Manufacturing in Pennsylvania have adopted cobots for welding and assembly, blending human oversight with automation. Startups such as Dyna Robotics are pushing further with AI-powered learning robots, backed by investors like Nvidia and Amazon, aiming to redefine industrial automation.
[ » Read full article *May Require Paid Registration ]
The Wall Street Journal; Christopher Mims (October 9, 2025)
Germany will vote against the EU’s “Chat Control” proposal, which would require tech companies to scan private digital messages for child sexual abuse material. Germany’s Federal Justice Minister Stefanie Hubig said such random chat monitoring violates democratic principles, stressing that “private communication must never be placed under general suspicion.” Despite Germany’s stance, countries like France and Spain continue to back the measure ahead of the scheduled Oct. 14 vote.
[ » Read full article ]
Computing (U.K.); Dev Kundaliya (October 9, 2025)

The subway advertisements were defaced An ad blitz for Friend.com, promoting a wearable AI “companion,” has inundated New York City’s subway walls, sparking debate and backlash. The minimalist ads, touting slogans like “I’ll never bail on our dinner plans,” have been widely defaced with anti-AI graffiti. Friend founder Avi Schiffmann urges people to consider AI as a new category of companionship that will coexist with, not replace, traditional friends.
[ » Read full article *May Require Paid Registration ]
The New York Times; Stefano Montali (October 7, 2025)
New Chair of the Department of Computer Science at Baylor University
 
ACM Insurance for Members
 

Association for Computing Machinery

1601 Broadway, 10th Floor
New York, NY 10019-7434
1-800-342-6626
(U.S./Canada)



ACM Media Sales

If you are interested in advertising in ACM TechNews or other ACM publications, please contact ACM Media Sales or (212) 626-0686, or visit ACM Media for more information.

To submit feedback about ACM TechNews, contact: [email protected]

Archives | Career News | Contact Us | Unsubscribe