This week’s list of top data news highlights covers July 28th, 2025 to August 3rd, 2025 and includes articles on training AI to detect brain tumors and assist disoriented pilots.
1. Designing Antivenom with AI
Researchers at the Technical University of Denmark and the University of Washington have developed an AI tool that designs synthetic proteins to neutralize snake venom. The model uses deep learning to identify antibodies that bind to specific toxins, and the candidates it proposed achieved high survival rates in mouse tests. Unlike traditional antivenoms, which are made by injecting animals with venom and harvesting their antibodies, this method avoids using animals in the production process, offering a faster and potentially more scalable alternative.
2. Spotting Weeds from the Sky
Researchers at Carnegie Mellon University and a Montana conservation ranch are using AI to help ranchers combat leafy spurge, an invasive weed that costs the U.S. beef and hay industry an estimated $35 million a year. The team is training image detection models to identify the plant from drone imagery. To overcome the shortage of real-world images, they use generative AI to create synthetic examples of leafy spurge under different weather and lighting conditions. This expanded dataset improves detection accuracy and reduces the need for on-the-ground fieldwork.
OffDeal, a New York-based AI investment bank, has developed a platform that automates nearly all of the administrative steps in selling small and mid-sized businesses. Once a company hires OffDeal to manage the sale, the platform handles key tasks typically done by a team of bankers. It scrapes data from millions of business websites to identify likely buyers, then uses generative AI to draft one-page teasers, track signed NDAs, and generate full company profiles for potential bidders, streamlining the entire process from outreach to auction.
4. Enforcing Teen Protections with AI
YouTube has developed an AI-powered tool that estimates user age by analyzing viewing history, video categories, and account age. It flags users who appear to be under 18, regardless of the birthdate listed on the account, and disables personalized ads and adjusts recommendations to reduce repeated exposure to content that may be inappropriate for younger audiences. Users who are incorrectly flagged can upload an ID to correct their age.
5. Combatting Spatial Disorientation
Researchers at the University of Maryland are developing a flight suit that uses vibrating sensors to help pilots recognize when they’re becoming disoriented during flight. Spatial disorientation causes 5 to 10 percent of general aviation accidents and is often fatal. The suit delivers directional vibrations, like a buzz on the left shoulder if the plane rolls too far right, providing an extra cue beyond visual instruments and physical sensation. In early simulator tests, pilots using the vibrations reacted faster and were better able to maintain control, especially in low-visibility conditions like clouds or night flying.
6. Shopping with a Stylist Bot
Daydream, a fashion-focused AI startup based in New York, has launched an AI-powered shopping tool that lets users search for clothing by describing what they want in natural language or uploading an inspirational photo. Its generative AI model interprets phrases like “a revenge dress for a party in Sicily” and returns product matches from over 8,000 brands. Users can then refine results through a chat-style interface, asking for cheaper options, different styles, or specific features. The system learns from browsing behavior over time to improve recommendations.
7. Tracking Wildfires in Real-Time
The Earth Fire Alliance, a coalition of government agencies, university researchers, and private industry partners, is building a wildfire monitoring system that uses satellites to detect new fires and drone swarms to track their spread on the ground. Satellites scan vast areas to spot early signs of ignition, while drones are deployed to map terrain, monitor fire behavior, and relay real-time data to responders. The system was put to use during the 2024 Fish Creek Fire in Wyoming, where it delivered faster and more detailed fire maps than traditional flyovers.
8. Talking Through Memory Loss
NewDays, an AI startup based in Seattle that focuses on cognitive health and aging, has developed an AI companion called Sunny for people with mild cognitive impairment or early-stage dementia. Users chat with Sunny on a computer, engaging in memory prompts, casual conversation, and cognitive exercises to keep the brain active. The system adapts over time by learning how and when users engage, allowing it to tailor prompts and initiate conversations based on individual patterns and needs.
9. Mapping Landslide Risk with AI
Italy’s environmental research institute, ISPRA, has launched an AI assistant to help residents assess landslide risk by guiding them through the IdroGEO platform, a public tool with maps and up-to-date data on land instability. The AI assistant answers user questions, helps locate specific areas on the map, and explains risk levels based on climate and geological data. The goal is to make it easier for people, especially those in high-risk zones, to access and understand detailed information about landslide threats.
10. Reading Between the Vitals
Google has introduced SensorLM, a suite of AI foundation models that turn raw data from wearables like heart rate and motion into natural-language summaries explaining why something happened, not just what. Unlike standard Fitbit features that show basic stats, SensorLM adds context, distinguishing between, say, “brisk uphill walk” or “public speaking stress,” based only on sensor data. Trained on nearly 60 million hours of real-world data, SensorLM is meant to power smarter health apps, digital coaches, and clinical tools.
Image credits: Pixabay
