Welcome to the February 3, 2025 edition of ACM TechNews, providing timely information for computer professionals three times a week.

HEADLINES AT A GLANCE

• Japan to Curb Exports of Chips, Quantum-Computing Tech
• Chinese-Made Patient Monitor Contains a Secret Backdoor
• International Police Op Takes Down Cybercrime Marketplaces
• Israeli Startups Make Global Plans After Key Role in War
• AI Systems with ‘Unacceptable Risk’ Now Banned in EU
• Asia Goes Cashless as Countries Push for Digital Payments
• U.S. Teens Increasingly Misled by Fake Content Online
• OpenAI to Provide Models to National Labs
• Chinese, Iranian Hackers Use U.S. AI Products to Bolster Cyberattacks
• 58% of Ransomware Victims Forced to Shut Down Operations
• DeepSeek's Chatbot Achieves 17% Accuracy in Audit

Japan to Curb Exports of Chips, Quantum-Computing Tech

The Japanese government is expanding its list of export-controlled items to include advanced chips, lithographic equipment, and cryocoolers needed for the manufacture of quantum computers, according to draft revisions to that nation’s foreign exchange law. Companies will need licenses to export those items to prevent their use in weapons or their development, said Economy Ministry officials. The new curbs are scheduled to go into effect at the end of May.
[ » Read full article ]

Chinese-Made Patient Monitor Contains a Secret Backdoor

The U.S. Cybersecurity and Infrastructure Security Agency (CISA) has warned of a threat discovered in three firmware versions of a patient monitor made by China-based Contec Medical Systems. The monitor was configured to connect to an IP address for a third-party university with no connection to the manufacturer, enabling the university to remotely download and execute unverified files on the patient monitor, CISA said. The backdoor also automatically sends patient data to the IP address.
[ » Read full article ]

International Police Op Takes Down Cybercrime Marketplaces

An international law enforcement operation has taken down two of the largest cybercrime marketplaces. In addition to the arrest of two individuals and confiscation of servers and other devices, the operation resulted in the shutdown of 12 accounts and two domains used by the cybercrime forums known as Cracked and Nulled.
[ » Read full article ]

Israeli Startups Make Global Plans After Key Role in War

Israeli startups are capitalizing on their wartime success, inking deals across the globe. For example, Xtend, which makes AI-powered drones able to perform high-precision strikes with minimal human involvement from 9,000 km (5,600 miles) away, has signed contracts with the U.S. government. Between October 2023 and the end of 2024, Israel's defense ministry awarded $219 million in contracts to 101 startups and small businesses to support the war effort.
[ » Read full article ]

AI Systems with ‘Unacceptable Risk’ Now Banned in EU

As of Sunday, EU regulators can ban the use of AI systems they deem to pose an “unacceptable risk” or harm under the bloc's AI Act, approved by the European Parliament last March. Unacceptable activities include the use of AI for social scoring, manipulating a person’s decisions deceptively, predicting people committing crimes based on their appearance, and trying to infer people’s emotions, among other uses.
[ » Read full article ]

Asia Goes Cashless as Countries Push for Digital Payments

QR codes and other smartphone-based technologies are rapidly replacing physical money as the method of payment in consumer transactions, driven in part by efforts from India and other countries to promote domestic digital settlement systems. In 2016, India introduced the Unified Payments Interface (UPI) mobile payment system that enabled real-time payments. More than 131 billion transactions were made via UPI in fiscal 2023.
[ » Read full article ]

U.S. Teens Increasingly Misled by Fake Content Online

A survey of 1,000 13- to 18-year-olds by the nonprofit advocacy group Common Sense Media revealed that around 35% of respondents had been deceived by fake online content; another 41% acknowledged having seen real yet misleading content, and 22% had shared information they later learned was false. Common Sense said respondents who came across fake online content were more likely to believe AI would make it harder to verify online information.
[ » Read full article ]

OpenAI to Provide Models to National Labs

OpenAI's o1 reasoning model, or another from its o-series, will be deployed on Los Alamos National Laboratory's Venado supercomputer. The deal with the U.S. government will make the model available to researchers at Lawrence Livermore and Sandia National Laboratories as well. Said Los Alamos' Thom Mason, "As threats to the nation become more complex and more pressing, we need new approaches and advanced technologies to preserve America's security."
[ » Read full article ]

Chinese, Iranian Hackers Use U.S. AI Products to Bolster Cyberattacks

Hackers linked to China, Iran, and other foreign governments are using the latest U.S. AI technology to bolster their cyberattacks, according to U.S. officials and security researchers. Google’s cyber-threat experts say that in the last year, dozens of hacking groups in more than 20 other countries deployed Google's Gemini chatbot to assist with malicious code writing and targeting.
[ » Read full article *May Require Paid Registration ]

58% of Ransomware Victims Forced to Shut Down Operations

A report from the Ponemon Institute found that 58% of organizations affected by ransomware attacks last year had to cease operations as a result, up from 45% in 2021. Forty percent of organizations experienced a substantial loss of revenue due to such an attack, up from 22%, while 35% reported brand damage, up from 21%.
[ » Read full article ]

DeepSeek's Chatbot Achieves 17% Accuracy in Audit

An audit by trustworthiness rating service NewsGuard found the chatbot rolled out by Chinese AI startup DeepSeek had an accuracy rate of 17% when it comes to delivering news and information. DeepSeek provided vague or useless answers 53% of the time and repeated false claims 30% of the time, with a fail rate of 83%. In comparison, its Western rivals, including OpenAI, had a 62% average fail rate.
[ » Read full article ]