Welcome to the August 6, 2025 edition of ACM TechNews, providing timely information for computer professionals three times a week.

HEADLINES AT A GLANCE

• How China Is Girding for an AI Battle with the U.S.
• Google AI Model Mines Trillions of Images to Create Maps of Earth ‘At Any Place and Time’
• Judge Allows the National Science Foundation to Withhold Hundreds of Millions of Research Dollars
• Thousands of ChatGPT Conversations Appearing in Google Search Results
• One-Fifth of Computer Science Papers May Include AI Content
• Flickering Lights Could Help Fight Misinformation
• Dutch Court Says Diesel Brands Now Owned by Stellantis Had Cheating Software from 2009
• Nearly Half of All Code Generated by AI Found to Contain Security Flaws
• Putin Widens Effort to Control Russia's Internet
• Palantir Gets $10-Billion Contract From U.S. Army
• Tesla Found Partly to Blame for Fatal Autopilot Crash
• China Says U.S Exploited Old Microsoft Flaw for Cyberattacks
• NIST Consortium and Draft Guidelines Aim to Improve Security in Software Development

How China Is Girding for an AI Battle with the U.S.

China is working to develop a self-sufficient AI ecosystem to counter U.S. export restrictions on advanced semiconductors. At Shanghai's World Artificial Intelligence Conference, companies showcased AI systems designed for Chinese-made chips. "Project Spare Tire," led by Huawei Technologies, is pushing for 70% semiconductor self-sufficiency by 2028 by clustering multiple domestic chips. China also unveiled an international open-source AI governance framework to challenge U.S. closed models.
[ » Read full article *May Require Paid Registration ]

Google AI Model Mines Trillions of Images to Create Maps of Earth ‘At Any Place and Time’

Google DeepMind’s AlphaEarth Foundations geospatial AI model can produce high-resolution Earth surface embeddings using satellite and environmental data captured between 2017 to 2024. The model enables spatial-temporal analysis for applications such as climate impact tracking, deforestation monitoring, and clean-energy site planning. It processes data into 10-meter resolution units and integrates it via Google Earth Engine. According to initial testing, AlphaEarth improves on existing models in data density and inference speed, although researchers caution further evaluation is needed to assess accuracy across use cases.
[ » Read full article *May Require Free Registration ]

Judge Allows the National Science Foundation to Withhold Hundreds of Millions of Research Dollars

On Aug. 1, a federal court declined to order the Trump administration to restore hundreds of millions of dollars in terminated funding that had been awarded to research institutions by the National Science Foundation. A coalition of 16 states argued that the cuts “violate the law and jeopardize America’s longstanding global leadership in STEM.” U.S. District Judge John Cronan in New York said he would not grant the preliminary injunction because the court may lack jurisdiction to hear the suit.
[ » Read full article ]

Thousands of ChatGPT Conversations Appearing in Google Search Results

Thousands of private ChatGPT conversations are appearing in Google search results, exposing deeply personal user disclosures. The issue stems from OpenAI’s shareable chat links, which included an optional, but often misunderstood, setting allowing conversations to be indexed by search engines. While the feature has since been removed, previously indexed chats remain public unless deleted by users. Some include details about trauma, mental health, or identity, raising concerns about data privacy, interface design, and broader industry responsibility around user protection and transparency.
[ » Read full article *May Require Free Registration ]

One-Fifth of Computer Science Papers May Include AI Content

Nearly one in five computer science papers published in 2024 may include AI-generated text, according to a large-scale analysis of over 1 million abstracts and introductions by researchers at Stanford University and the University of California, Santa Barbara. The study found that by September 2024, 22.5% of computer science papers showed signs of input from large language models like ChatGPT. The researchers used statistical modeling to detect common word patterns linked to AI writing.
[ » Read full article ]

Flickering Lights Could Help Fight Misinformation

A team of Cornell University computer scientists developed "noise-coded illumination," a watermarking technique that embeds verification data as imperceptible flickers of light sources in videos. Cameras record these unique flicker patterns, enabling investigators to decode them and verify video authenticity or detect manipulations. Studio lights could be programmed to flicker in unique patterns during a recording, which would then be detectable by a decoder to determine if the video was later altered.
[ » Read full article ]

Dutch Court Says Diesel Brands Now Owned by Stellantis Had Cheating Software from 2009

Diesel cars sold in the Netherlands by Opel, Peugeot, Citroen, and DS since 2014, and likely since 2009, were equipped with software that manipulated their emission control systems to cheat emissions tests, according to a July 30 Dutch court ruling in a class action lawsuit against Stellantis, owner of the automobile companies. The court said the software was designed to maintain artificially low levels of nitrogen oxide emissions during official tests. Stellantis denied the accusations.
[ » Read full article ]

Nearly Half of All Code Generated by AI Found to Contain Security Flaws

New research from application security solution provider Veracode reveals that 45% of all AI-generated code contains security vulnerabilities, with no clear improvement across larger or newer large language models. An analysis of over 100 models across 80 coding tasks found Java code most affected with over 70% failure, followed by Python, C#, and JavaScript. The study warns that increased reliance on AI coding without defined security parameters, referred to as "vibe coding," may amplify risks.
[ » Read full article ]

Putin Widens Effort to Control Russia's Internet

Russian President Vladimir V. Putin signed new laws to further restrict online freedom in that nation, which could impede the use of YouTube and other services from U.S. technology companies. The government is creating a state-approved messaging service, MAX, which will be pre-installed on all smartphones sold in Russia beginning in September. MAX could become Russia's version of WeChat, which China's residents use to message each other, pay bills, and apply for marriage licenses, among other tasks.
[ » Read full article *May Require Paid Registration ]

Palantir Gets $10-Billion Contract From U.S. Army

The U.S. Army awarded Palantir a contract worth up to $10 billion over the next 10 years, the largest in the company’s history. This agreement signifies a major shift in the Army’s software procurement approach by consolidating existing contracts to achieve cost efficiencies and expedite soldiers' access to advanced data integration, analytics, and AI tools. The contract aligns with the Pentagon's strategic focus on enhancing data-mining and AI capabilities amid escalating global security challenges.
[ » Read full article *May Require Paid Registration ]

Tesla Found Partly to Blame for Fatal Autopilot Crash

A Florida jury on Aug. 1 found that flaws in Tesla’s self-driving software were partly to blame for a 2019 crash that killed a 22-year-old woman and severely injured another. The verdict is a significant setback for the carmaker, which is staking much of its future on developing self-driving taxis. If upheld on appeal, the verdict would require Tesla to pay as much as $243 million in punitive and compensatory damages.
[ » Read full article ]

China Says U.S Exploited Old Microsoft Flaw for Cyberattacks

The Cyber Security Association of China has accused U.S. hackers of stealing military data and perpetrating cyberattacks against the nation's defense sector. The association said the U.S. actors exploited vulnerabilities in Microsoft Exchange email servers to attack two major Chinese military companies, which it did not name. The hackers reportedly controlled the servers of one key defense company for almost a year, according to the association.
[ » Read full article *May Require Paid Registration ]

NIST Consortium and Draft Guidelines Aim to Improve Security in Software Development

The National Institute of Standards and Technology's (NIST) National Cybersecurity Center of Excellence (NCCoE), together with 14 member organizations in its Software Supply Chain and DevOps Security Practices Consortium, is developing guidelines for secure software development in response to White House Executive Order 14306. Their draft, NIST Special Publication 1800-44, outlines high-level DevSecOps practices and intends to expand on the Secure Software Development Framework (SSDF). Public comments on the guidelines are being accepted until September 12, 2025.
[ » Read full article ]