Apple and medical AI company OpenEvidence are studying how people use and trust AI health advice through the Apple Health AI Usage Study app, announced at the HLTH conference in Las Vegas on November 15, 2026. Apple plans to recruit up to 200,000 eligible U.S. participants for a year-long study with the University of California, San Francisco, using surveys, interviews and data analysis. Eligible users must be at least 21 and have an iPhone 15 Pro or later running iOS 26 or later. The research will examine what people ask, how well they understand responses, what builds trust and whether AI changes health behavior or helps users prepare for doctor visits. Apple health and fitness vice president Sumbul Desai said the work could inform future health products; OpenEvidence CEO Daniel Nadler described the partnership as bringing together brands trusted by doctors and consumers.
The experimental app answers health questions using OpenEvidence’s medical AI and citations from peer-reviewed research, including The New England Journal of Medicine and the Journal of the American Medical Association. Users can rate answers, complete surveys and optionally share phone health data for more tailored responses. Apple says identifying details are removed before questions reach OpenEvidence, which will delete questions and answers within 24 hours and not use them to train its model. The app can flag concerning questions and direct users to care, but it does not provide live medical monitoring or replace a clinician. Apple says the findings will help it present complex health information responsibly.
Meanwhile, a separate infrastructure analysis argues that businesses will increasingly combine hyperscale data centers with edge computing, which processes information closer to where it is generated. The choice depends on speed, cost, security and control. Five major technology companies spent more than $400 billion on capital projects in 2025; data centers used 485 TWh of electricity globally, a figure projected to reach 950 TWh by 2030. Power constraints, expense and security concerns are driving interest in hybrid systems, while the global edge data center market is projected to grow 22% annually over the next decade.
AI adoption and security risks are also rising in manufacturing. Netskope Threat Labs reported on November 16, 2026, that 67% of manufacturing users actively use AI tools—double the share a year earlier—and Microsoft Copilot is the sector’s most widely used AI application. The report says 96% of employees use applications with built-in AI features, while regulated data accounted for 57% of AI-related data policy violations. Malicious AI-link interactions reached 200 per 100,000 users in March 2026. Netskope recommends inspecting HTTP and HTTPS downloads and using data loss prevention policies and Remote Browser Isolation.
Other reports show AI moving into distinct practical roles. Commerce.com launched an AI Shopping Agent for BigCommerce merchants, offering personalized product recommendations, real-time answers and checkout guidance. SingNova, a company incubated by Nanyang Technological University, showcased RISC-V-based AI computing, virtual employees and intelligent video storage at Safety & Security Asia 2026 in Singapore. In chip design, AI agents could coordinate tools and explore workload-specific designs, though engineers still need to verify results; adoption raises questions about trust, security, permissions and proprietary data.
Research and legal cases illustrate the limits of AI outputs. Finite State will present six months of AI-assisted firmware security research at Hardwear.io 2026: its pipeline flagged 802 possible remote code execution vulnerabilities, but researchers demonstrated 68 with reproducible proofs; more than 140 exploits were demonstrated when memory-corruption cases were included. In Arizona, Commissioner Daniel Fenzel released DUI suspect Anthony Edward Rene Quidera after saying he could not determine whether factual claims in a police report were written by the officer or generative AI. The report disclosed Truleo AI assistance; Quidera later failed to appear for his November 12 trial, according to ABC15 Arizona.
Researchers also tested AI-generated food: trained on more than 2,200 human-written recipes, a system produced five burgers compared with a Big Mac in a blind tasting of 101 people. Some AI recipes scored slightly higher on flavor, while mushroom- and bean-based options performed better on environmental or nutritional qualities but generally rated lower for taste. Futurist Thomas Frey, meanwhile, argues that AI could move computing beyond visible interfaces toward interactions among people, agents and ambient systems, making discretion, timing and trust important to future device design.
Key Takeaways
- Apple and OpenEvidence launched a year-long study of conversational health AI, with Apple aiming to enroll up to 200,000 eligible U.S. participants alongside UCSF.
- The Apple study examines trust, comprehension, everyday use and potential effects on health behavior; Sumbul Desai and OpenEvidence CEO Daniel Nadler described its goals and partnership.
- The study app uses research citations, offers optional phone health-data sharing and safety referrals, and has stated privacy protections; it is not a substitute for clinicians or live monitoring.
- Five major tech companies spent more than $400 billion on capital projects in 2025; global data-center electricity use is projected to rise from 485 TWh to 950 TWh by 2030, strengthening the case for hybrid cloud-and-edge infrastructure.
- In manufacturing, 67% of users actively use AI tools, Microsoft Copilot leads sector adoption, and regulated data accounted for 57% of AI-related policy violations, according to Netskope.
- Commerce.com introduced an AI Shopping Agent for BigCommerce merchants, while SingNova showcased AI computing, virtual employee and video-storage solutions in Singapore.
- AI agents may support custom chip design, but engineering verification and concerns about security, permissions, trust and proprietary data remain.
- Finite State’s firmware pipeline flagged 802 possible remote code execution vulnerabilities; researchers demonstrated 68 with reproducible proofs and more than 140 exploits including memory-corruption cases.
- An Arizona commissioner released a DUI suspect after questioning whether an AI-assisted police report’s factual claims could be attributed to the officer; the report disclosed Truleo AI assistance.
- Blind taste tests found some AI-designed burgers rated slightly higher for flavor, while mushroom- and bean-based options scored better on environmental or nutritional qualities but generally lower on taste.
Apple tests a health chatbot in new research study
Apple launched the Apple Health AI Usage Study app to learn how people use and understand AI health advice. The app works with OpenEvidence, a medical AI tool, and can provide answers with links to peer-reviewed research. Participants may choose to share health data from their phones for more personal responses, and they can rate answers and complete surveys. Apple says privacy steps remove identifying details before questions reach OpenEvidence, which will delete question and response data within 24 hours and will not use them to train its model. The app includes safety guidance, but it does not provide live medical monitoring or replace a healthcare professional.
Apple and OpenEvidence launch study of conversational health AI
Apple and OpenEvidence announced a health AI study and app at the HLTH conference in Las Vegas on November 15, 2026. The study will examine user trust, how well people understand answers, response quality, and whether AI can affect health behavior. Apple designed the study, while OpenEvidence supplies the AI; its medical search tool is used by nearly two-thirds of U.S. physicians, the company says. Apple hopes the findings will help it develop healthcare AI products and present complex information more clearly. The app can also flag concerning questions and direct users to outside care or other resources.
Apple plans study of 200,000 people using health AI
Apple launched the Apple Health AI Usage Study to learn how people use conversational health AI and how it may affect their health. The company plans to recruit 200,000 participants in the United States. The study will examine what health information people seek and how they interact with AI tools. It will be conducted with the University of California, San Francisco, using surveys, interviews, and data analysis. Apple says the results will help guide the development of health AI tools.
Apple and OpenEvidence test an AI health chat app
Apple and OpenEvidence launched a study on Sunday in Las Vegas using an experimental AI-powered health chat app. At the HLTH conference, Apple health and fitness vice president Sumbul Desai said the study will examine how consumers use health chatbots, including what builds trust and how well people understand responses. She said Apple chose OpenEvidence as a partner because it is clinically grounded. OpenEvidence CEO Daniel Nadler said the companies bring together trusted brands for doctors and consumers.
Apple study will examine how people use AI for health advice
Apple is working with OpenEvidence on an app-based study of how people use AI to answer health questions. The study will examine trust, understanding, and everyday use, and may explore whether AI helps people prepare for doctor visits. The app can answer basic and more complex questions, drawing on OpenEvidence citations from peer-reviewed journals such as The New England Journal of Medicine and the Journal of the American Medical Association. Apple says users with concerning responses may be advised to contact a healthcare provider. The study is open to up to 200,000 eligible participants nationwide who are at least 21 and have an iPhone 15 Pro or later running iOS 26 or later.
Apple studies how people use and trust AI health chatbots
Apple has launched a year-long project called the Apple Health AI Usage Study. It will examine how people use health chatbots and how much they trust their advice. The study comes as people continue using chatbots even though they can give dangerous or incorrect answers. Apple says the research aims to help make AI health advice more responsible.
AI infrastructure shifts toward a balance of hyperscale and edge
Companies are weighing large central data centers against edge computing, which processes data closer to where it is created. The article argues that businesses will need hybrid systems shaped by speed, cost, security and control, rather than choosing only one approach. Five major tech companies spent more than US$400 billion on capital projects in 2025, while global data center electricity use reached 485 TWh and is expected to rise to 950 TWh by 2030. Power shortages, rising costs and security risks are adding pressure to infrastructure decisions. The global edge data center market is projected to grow 22% each year over the next decade.
Arizona judge releases DUI suspect after AI use in police report
An Arizona judge released Anthony Edward Rene Quidera after finding that the court could not determine probable cause in his DUI case. Quidera was arrested on Oct. 29 after allegedly driving his Dodge Charger over 100 mph, crossing lane markings and having an open container; the trooper also reported smelling alcohol. The arrest report disclosed that Truleo AI helped generate it, while stating that the facts came from the officer’s own observations and recollections. Commissioner Daniel Fenzel said he could not meaningfully assess whether the report’s factual claims were written by the officer or generative AI. Quidera was released but did not appear for his Nov. 12 trial, according to ABC15 Arizona.
AI-designed burgers put to the test against the Big Mac
Researchers trained generative AI on more than 2,200 human-written burger recipes to create burgers aimed at improving taste, nutrition or environmental sustainability. They compared five AI-designed burgers with a Big Mac in a blind tasting with 101 participants. Some AI burgers received slightly higher flavor ratings. Mushroom- and bean-based options scored better for environmental or nutritional qualities, but generally received lower sensory ratings.
Finite State to share results from AI firmware security research
Finite State will present findings from six months of AI-assisted firmware security research at Hardwear.io 2026. The team examined 35 firmware images, while the wider research group analyzed more than 50 across routers, industrial controllers, automotive systems and other devices. Its AI-driven pipeline flagged 802 possible remote code execution vulnerabilities, but researchers demonstrated only 68 through reproducible proofs; more than 140 exploits were demonstrated when memory-corruption cases were included. The session will explain why flagged vulnerabilities are not always exploitable and cover the team’s methods, limits, costs and false positives. Roland Lindsey, Finite State’s Chief Product Security Engineer, is listed as the speaker.
AI Agents Could Become Reusable Tools for Chip Design
AI agents could make custom chip design workflows faster and less costly to build. Unlike traditional semiconductor intellectual property, agents can explore design choices, coordinate tools, and customize functions for different workloads, though engineering teams still need to check their results. Multi-agent systems also create challenges in coordinating tasks, verifying work, and connecting tools. New pricing models could change the electronic design automation business, but adoption faces concerns about trust, security, permissions, and protecting proprietary data.
Manufacturing AI Use Doubles as Security and Data Risks Grow
Netskope Threat Labs reported on November 16, 2026, that 67% of manufacturing users actively use AI tools, twice the share from a year earlier. Microsoft Copilot is the sector’s most widely used AI application, while 96% of employees use applications with built-in AI features. Regulated data made up 57% of AI-related data policy violations, and malicious AI link interactions reached 200 per 100,000 users in March 2026. Netskope recommends checking HTTP and HTTPS downloads and using data loss prevention policies and Remote Browser Isolation to help protect networks.
Commerce.com Launches AI Shopping Agent for BigCommerce Merchants
Commerce.com introduced an AI-powered Shopping Agent for merchants using the BigCommerce platform. The tool is designed to help customers find products through personalized recommendations, get answers in real time, and receive guidance during checkout. Commerce.com says the feature aims to make shopping easier and potentially increase sales for its partners.
SingNova Presents AI and Video Storage Solutions in Singapore
SingNova, a high-tech company incubated by Nanyang Technological University in Singapore, is showcasing its portfolio at Safety & Security Asia 2026. The event takes place at Marina Bay Sands in Singapore from November 10 to 12. SingNova’s offerings include RISC-V-based AI computing, AI virtual employee solutions, and intelligent video storage.
AI Could Make Interfaces Disappear and Change How People Use Technology
Futurist Thomas Frey describes a new shift in how people interact with computers. Earlier changes moved from typing commands to using a mouse and then tapping screens, but this shift involves machines learning to understand people. Frey says design will move beyond visible interfaces toward relationships among people, AI agents, and ambient systems. He argues that companies will need restraint, discretion, good timing, and trust to succeed with future AI devices.
Sources
- Apple launches an AI health chatbot as part of a research study
- Apple, OpenEvidence launch conversational health AI study and app
- HLTH26: Apple launches 200K-person study to examine how consumers use health AI
- Apple and OpenEvidence launch experimental health chat app
- Apple plans to study how consumers use AI for health needs: Questions and answers
- New Apple health study wants to find out how much you trust AI
- Hyperscale vs Edge: The New Rules of AI Infrastructure
- Judge releases DUI suspect after reading AI-assisted police report disclosure
- Can AI make a better burger than a Big Mac? Scientists put it to the test
- Finite State to Reveal Results of Six-Month, AI-Driven Firmware Security Research at Hardwear.io 2026
- Are AI Agents The New Semiconductor IP?
- Manufacturing AI adoption doubles as data governance gaps, shadow AI and emerging cybersecurity threats rise, Netskope finds
- Commerce.com unveils AI assistant to boost sales for BigCommerce merchants
- SingNova Showcases RISC-V-Based AI Computing and Intelligent Video Storage Portfolio at Safety & Security Asia 2026
- The Vanishing Interface: How AI Will Change Everything We Touch, Say, and See
Comments
Please log in to post a comment.