A group of top Silicon Valley executives—including Meta CTO Andrew Bosworth, OpenAI’s Kevin Weil, and Palantir CTO Shyam Sankar—have been inducted into the U.S. Army Reserve as part of a new elite unit called Detachment 201: the Executive Innovation Corps. Bypassing basic training, these newly minted lieutenant colonels will remotely advise the military on applying cutting-edge tech for combat efficiency and modernization, while keeping their civilian jobs. Spearheaded by Pentagon talent chief Brynt Parmeter and fast-tracked amid shifting attitudes in the Valley, the program aims to infuse military strategy with private-sector innovation. Click here to read more.

As AI becomes more embedded in daily life, concerns over its energy use and environmental impact are growing—but reliable data remains elusive. OpenAI CEO Sam Altman claimed that a ChatGPT query uses 0.34 watt-hours, yet experts argue this lacks transparency and likely omits factors like model training and server cooling. A study shows that most AI usage involves models with no environmental disclosures, and myths—like ChatGPT being 10 times more energy-intensive than a Google search—often stem from unverified remarks. Researchers urge better disclosure and smarter deployment of low-energy models to curb emissions, as current practices obscure AI’s true carbon cost. Click here to read more.

Google is testing Search Live, a new feature that allows real-time voice conversations with its Gemini AI directly within the Google Search app on Android and iOS. Currently available to Labs users in the U.S., the tool lets users ask spoken questions, receive audio responses, and view relevant links—all while continuing conversations across apps. Future updates will include camera integration, allowing users to point at objects and ask questions out loud. Users can also view transcripts or respond by typing, with conversations saved in their AI Mode history. The move positions Google alongside rivals like OpenAI and Anthropic in the race to build seamless AI voice interfaces. Click here to read more.

After vetoing California’s ambitious AI safety bill last year, Governor Gavin Newsom tasked experts with drafting an alternative—and now their new report proposes a sweeping framework to regulate powerful AI models. The “California Report on Frontier Policy” outlines 19 key recommendations, including third-party safety testing, whistleblower protections, and transparency mandates for AI developers. As Congress weighs a decade-long moratorium on state-level AI laws, the report pushes back, arguing California can lead in crafting smart, flexible oversight amid the risks of fast-evolving technologies. Click here to read more.

While Apple’s WWDC keynote didn’t spotlight smart home updates, a new developer video quietly unveiled a major expansion: Apple Home is moving toward full energy management. With the introduction of EnergyKit in iOS 26 and iPadOS 26, developers can now access real-time energy data—like usage, rates, and clean energy forecasts—to help devices like thermostats and EV chargers operate more efficiently and cost-effectively. Though still limited, this positions Apple to eventually transform its Home app into a full-fledged home energy management system, signaling a bold shift toward smarter, greener living. Click here to read more.

As AI becomes more embedded in software development, many programmers find themselves proofreading machine-generated code more often than writing their own—a shift some say makes their jobs feel less creative and more frustrating. While tools like GitHub Copilot and Cursor can speed up boilerplate tasks, developers warn that AI code still needs thorough review, similar to mentoring a junior colleague. Despite management pressure to adopt AI for speed and efficiency, most developers agree it lacks the nuanced understanding of complex systems, requiring oversight to maintain quality and long-term maintainability. Click here to read more.

While major AI companies claim that respecting copyright is impractical when building large language models, a team of researchers has proven otherwise—though not without serious effort. In a new paper, scientists from institutions like MIT and Eleuther AI built an 8-terabyte dataset using only public domain and openly licensed text, training a model comparable in performance to Meta’s Llama 2-7B. The process, however, required extensive human labor to ensure legal compliance and data quality, highlighting the tradeoff between ethics and efficiency. Their work challenges the tech industry’s claims and could reshape the debate over AI, copyright, and transparency. Click here to read more.

Walmart is dramatically expanding its drone delivery program through a partnership with Alphabet’s Wing, planning to add the service to 100 more stores in cities like Atlanta, Charlotte, and Houston over the next year. Promising deliveries in under 30 minutes, the initiative will become the largest drone network in the U.S., offering free or discounted service to Walmart+ members. While the technology has seen success in select markets like Dallas, experts remain skeptical about its long-term profitability due to regulatory, logistical, and economic challenges. Click here to read more.