Google Gemini drafts rules as Amazon launches Bee

Key Takeaways

• The US Department of Transportation is using Google Gemini to draft federal safety rules for transportation, with former President Donald Trump reportedly supporting the initiative to speed up rulemaking.
• Concerns exist among staffers and experts regarding potential AI errors and fabricated information when using AI for complex regulatory drafting.
• Amazon's new AI wearable, Bee, priced at $49.99, records thoughts and conversations, raising privacy concerns due to its ability to record voices without explicit permission, similar to past issues with Alexa.
• North Korean hacking group Konni is utilizing AI-generated PowerShell malware in spear-phishing campaigns targeting blockchain developers in Japan, Australia, and India.
• Mobileye has developed a new AI architecture to improve robotaxi economics by reducing the need for human remote operators, planning to deploy 100,000 robotaxis with Volkswagen by 2033.
• Momei Qu, a Managing Director at PSP Growth, states that companies prioritizing the execution of artificial intelligence are in a strong market position.
• Many organizations fail to realize value from AI in ERP systems due to inadequate redesign of operational processes and insufficient employee training on AI insights.
• New research from MIT suggests that less data can sometimes lead to better AI decisions, potentially enhancing transparency and reducing costs in data-intensive fields.
• The European Union is unlikely to experience an AI investment boom similar to the United States, facing challenges such as limited risky capital and infrastructure.
• The EU Chips Act aims to mobilize 86 billion euros by 2030 for microchip manufacturing, a scale still behind industrial policies in the US and China.

US Transportation Department uses AI for safety rules

The US Department of Transportation (DOT) is using artificial intelligence, specifically Google Gemini, to draft safety rules for airplanes, cars, and pipelines. A ProPublica investigation revealed concerns from staffers and experts about potential AI errors and fabricated information. DOT's top lawyer, Gregory Zerzan, believes AI can speed up the rulemaking process, aiming for drafts in 30 days or less. Staffers are skeptical, noting the complex nature of rule-making and the risks of flawed laws. Despite these worries, DOT has already used Gemini for an unpublished Federal Aviation Administration rule, with former President Donald Trump reportedly supporting the initiative.

Trump administration plans AI for federal rules

The Trump administration plans to use artificial intelligence, specifically Google Gemini, to draft federal regulations. The Department of Transportation (DOT) will lead this effort, aiming to create new rules for transportation safety, including airplanes, cars, and pipelines. Gregory Zerzan, DOT's top lawyer, stated the goal is to speed up the process, with Gemini drafting rules in as little as 20 minutes. Former President Donald Trump is reportedly very excited about this initiative, seeing DOT as a leader for other agencies. However, critics like former DOT AI officer Mike Horton warn that relying on AI for complex safety regulations could lead to errors and harm.

Konni hackers use AI malware on blockchain developers

The North Korean hacking group Konni is using AI-generated PowerShell malware to attack blockchain developers and engineering teams. Their phishing campaigns now target Japan, Australia, and India, expanding beyond their usual focus on South Korea. The hackers send spear-phishing emails with malicious links disguised as ad URLs to bypass security. These emails trick recipients into downloading ZIP files from WordPress sites, which contain a Windows shortcut that installs a remote access trojan called EndRAT. A newer campaign documented by Check Point uses Discord's content delivery network to host ZIP files that deploy a multi-stage attack, including a PowerShell backdoor and a legitimate remote monitoring tool, SimpleHelp.

January 2026 AI advances in radiology

This report, "Advances in AI January 2026," summarizes the latest news and research related to artificial intelligence in radiology. Diagnostic Imaging's monthly roundup helps professionals catch up on new AI developments. It covers AI-powered imaging tools and emerging research that impacts the field of radiology. Readers can review a slideshow to see the highlights from the past month.

AI focused companies are well positioned says Qu

Momei Qu, a Managing Director at PSP Growth, believes that companies prioritizing the execution of artificial intelligence are in a strong market position. She shared her insights on Bloomberg's "Surveillance" program with Jonathan Ferro and Annmarie Hordern. This discussion took place on January 26, 2026, ahead of a significant week for tech earnings.

Amazon Bee AI gadget raises privacy concerns

Amazon's new AI wearable gadget, Bee, allows users to record thoughts, events, and conversations, then access written transcripts in an app. While the company claims strong privacy protections, experts and users worry about potential issues. The Bee device, which costs $49.99, can record voices of people nearby without their clear permission. Concerns also arise from Amazon's past privacy problems with products like Alexa and Ring, including human reviewers listening to recordings and fines for keeping children's data. The Bee's co-founder, Maria de Lourdes Zollo, described the device as building a rich picture of a user's life, fueling fears that it might be constantly listening.

Mobileye AI architecture improves robotaxi economics

Mobileye has developed a new AI architecture that could significantly improve the economics of robotaxis. Currently, robotaxis require many remote operators to handle situations the onboard AI cannot, making them expensive. Mobileye's "fast-think, slow-think" system splits AI tasks, keeping critical safety decisions in the vehicle for speed and offloading less urgent, complex reasoning to the cloud. This approach aims to drastically reduce the need for human remote operators, which Goldman Sachs estimates currently requires one operator for every three cars. Mobileye is partnering with Volkswagen to deploy 100,000 robotaxis by 2033, starting with removing safety drivers this year.

Why AI fails to deliver value in ERP systems

Many organizations invest heavily in artificial intelligence within their Enterprise Resource Planning (ERP) systems but fail to see expected results. The main issue is not the AI technology itself, but how people and systems interact with its insights. Often, AI recommendations are reviewed, debated, or ignored, causing value to leak at the "last mile" of decision-making. This problem stems from organizations adopting AI without redesigning their operational processes, decision-making structures, or training their workforce. Employees often lack the experience and guidance to effectively use AI tools, leading them to revert to old methods. To truly benefit from AI in ERP, companies must rethink how work is done and integrate AI as a core driver of change, not just an added feature.

Less data can improve AI decisions say MIT

For a long time, experts believed that more data always led to better artificial intelligence decisions. However, new research from MIT challenges this idea, suggesting that more data is not always better. Researchers asked what the smallest amount of data is needed to make the best decision. This work has big implications for industries like finance, where showing an optimal decision based on minimal data can boost transparency and governance. Reducing data needs can also cut costs for storage and processing, and lower risks related to data privacy. This new approach could change how AI systems are designed in fields where collecting data is expensive or limited.

EU unlikely to see US style AI investment boom

The European Union is unlikely to experience an artificial intelligence investment boom similar to the United States, which significantly boosts US economic growth. Experts believe the EU's AI sector is too small and relies too much on imports to drive near-term GDP growth. Challenges like limited risky capital, long planning processes, and a lack of cheap, reliable electricity and cooling hinder AI investment in the EU. This situation is expected to widen the gap between the EU and the US in AI development. While the EU Chips Act aims to mobilize 86 billion euros by 2030 for microchip manufacturing, it still falls behind the scale of industrial policies in the US and China.

Sources

• “Wildly irresponsible”: DOT's use of AI to draft safety rules sparks concerns
• Government by AI? Trump Administration Plans to Write Regulations Using Artificial Intelligence
• Konni Hackers Deploy AI-Generated PowerShell Backdoor Against Blockchain Developers
• Advances in AI — January 2026
• Companies Focused on AI Execution Well-Positioned: Qu
• Amazon's AI 'Bee' Gadget Sounds Like A Privacy Nightmare - Here's Why
• Mobileye's New AI Architecture Could Solve a Problem Crushing Robotaxi Economics
• AI’s Last-Mile Failure: Why ERP Value Breaks Down After the Investment
• More Data Isn’t Always Better for AI Decisions
• No US-style AI investment boom to drive EU growth