ACM TechNews, Monday, August 26, 2024

Welcome to the August 26, 2024 edition of ACM TechNews, providing timely information for computer professionals three times a week.

AI Could Help Shrinking Pool of Coders Keep Outdated Programs Working

An AI model developed by researchers at Vietnam's FPT Software AI Center could allow COBOL-based systems to remain operational as the number of engineers familiar with the older programming language continues to decline. The researchers are training the XMainframe model to interpret COBOL code and rewrite it in other programming languages. In tests, the model outperformed other AI models in accurately summarizing the purpose of COBOL code.
[ » Read full article *May Require Paid Registration ]

Protecting Connected Self-Driving Vehicles from Hackers

University of Michigan (U-M) researchers found that connected self-driving vehicles are vulnerable to data fabrication attacks, which occur when hackers remove real objects from or insert fake objects into perception data. Researchers at U-M's Mcity Test Facility used falsified LiDAR-based 3D sensor data and zero-delay attack scheduling to better understand the security vulnerabilities, and developed the Collaborative Anomaly Detection system as a countermeasure. The system uses shared 2D occupancy maps to cross-check the data and quickly identify geometric inconsistencies.
[ » Read full article ]

Machine Learning Algorithm Improves 3D Printing Efficiency

Washington State University (WSU) researchers developed a machine learning algorithm that identifies the most efficient 3D print settings for producing complex structures. The researchers used the algorithm to optimize the design for kidney and prostate organ models, with a focus on geometric precision, weight, porousness, and printing time. WSU's Eric Chen said, "We were able to strike a favorable balance and achieve the best possible printing of a quality object, regardless of the printing type or material shape."
[ » Read full article ]

Toward a Code-Breaking Quantum Computer

Massachusetts Institute of Technology (MIT) researchers have developed an algorithm that could help pave the way for encryption methods strong enough to withstand a quantum computer's code-breaking power and feasible to implement. The new algorithm uses a series of Fibonacci numbers requiring simple multiplication instead of squaring, which allows any exponent to be computed using only two qubits. It also addresses error correction, filtering out corrupt results and processing only correct ones.
[ » Read full article ]

Brainwave-Reading Robot 'Coaches' Could Rehabilitate Stroke Survivors

Researchers at the U.K.'s National Robotarium and the AIT Austrian Institute of Technology have developed a robotic rehabilitation system that could help patients recovering from strokes and brain injuries regain the use of their limbs. The system’s headset detects neural activity via a brain computer interface, which enables the robot to understand intended movements. The robots are programmed as rehabilitation coaches to help patients complete personalized rehabilitation routines without human intervention.
[ » Read full article ]

AI Researchers Call for 'Personhood Credentials' as Bots Get Smarter

A team including researchers from OpenAI, Microsoft, and Harvard University have proposed the development of "personhood credentials" to help distinguish humans from bots online. Such a system would require humans to verify their identities offline to receive an encrypted credential allowing them to access an array of online services. The researchers proposed multiple personhood credentialing systems be created so users have options, and a single entity does not control the market.
[ » Read full article ]

The Washington Post; Will Oremus (August 21, 2024)

Google Joins $250-Million Deal to Support Newsrooms in California

A first-of-its-kind agreement announced by Google, the California News Publishers Association, and California lawmakers will provide newsrooms in the state with upwards of $250 million, while allowing Google to avoid proposed legislation requiring tech companies to pay news organizations when advertising appears alongside articles on the tech company's platform. Under the five-year deal, a New Transformation Fund would be created to preserve and expand journalism in the state.
[ » Read full article *May Require Paid Registration ]

The Citation Black Market: Schemes Selling Fake References Alarm Scientists

There are growing concerns about a black market that allows scientists to purchase bogus citations to pad their Google Scholar profiles. New York University computer scientists conducted a sting operation by purchasing 50 citations for $300 and adding them to a fake Google Scholar profile they created. The researchers also proposed a citation-concentration index to identify scientists with a large number of citations from only a few sources.
[ » Read full article ]

The 50-Year-Old Technology Brands and Retailers Are Trying to Quit

Brands and retailers could replace barcodes with QR codes as early as 2027, but such a transition will require retailers to update the software and hardware they use, while brands will need to make manufacturing changes, and collaboration will be required across industries worldwide. Further, customers used to scanning barcodes at self-checkout kiosks would have to be educated on how to interact with QR codes. The goal is to use a single QR code for point-of-sale and inventory scans, as well as customer engagement.
[ » Read full article *May Require Paid Registration ]

'Styx Stealer' Malware Developer Accidentally Exposes Personal Info to Researchers in 'Critical Opsec Error'

Check Point security researchers reported a "significant operational security error" by the suspected developer of the new Styx Stealer malware in which he exposed his identity and "provided valuable intelligence about other cybercriminals." The information leak occurred when the Styx Stealer developer used a Telegram bot token provided by a customer involved in the Agent Tesla malware campaign to debug the stealer on his own computer. The leaked information included his Telegram accounts, emails, and contacts.
[ » Read full article ]

Multiple Flaws in Microsoft macOS Apps Unpatched Despite Potential Risks

Security researchers at Cisco Talos identified eight flaws in Microsoft's macOS apps that could allow hackers to access a device to record video and sound, obtain sensitive data, log user input, and escalate privileges. The vulnerabilities affect Microsoft products Excel, OneNote, Outlook, PowerPoint, Teams, and Word. The researchers said Microsoft considers the flaws to be low risk and has no plans to fix them.
[ » Read full article ]