This is today’s edition of The Download, our weekday newsletter that provides a daily dose of what’s going on in the world of technology. The US spent billions on border surveillance. Why can’t it catch people before they die? When José Morales Bernal crossed the border into the US in April 2024, the day before his 32nd birthday, he was within range of three surveillance towers equipped with cameras and AI to automatically detect and track people. If the system worked as intended, Morales should have been apprehended. If he needed medical help, agents were trained to provide it. None of that happened. Instead, Morales died just 360 feet from the closest tower. It was local landfill workers, rather than Border Patrol, who first spotted him. An autopsy concluded that he had died of “environmental exposure.” A first-of-its-kind investigation by MIT Technology Review reveals that deaths like Morales’s are startlingly common. After mapping nearly 4,000 locations where human remains were found against information on nearly 600 surveillance towers, we discovered that a humanitarian crisis at the border has unfolded in view of the government’s own cameras. Read our full investigation into the deadly failings of the virtual wall. —James O’Donnell and Eileen Guo Learn more: As part of our 15-month investigation, we created the first comprehensive map of deaths near US-Mexico border surveillance towers. Take a look at that map, and read about how we made it. The US government is about to spend another $1 billion to triple the virtual wall’s size, yet we found it has systemic flaws: broken towers, algorithms that failed to detect people and agents who simply did not respond to alerts. Here are four ideas for what Customs and Border Protection should do to rectify those failures. An analysis by the Times of San Diego found at least 138 cases since 2022 in which migrants’ remains were found within the nominal range of a nearby surveillance tower. A separate MIT Technology Review analysis estimated that more than half of those people were likely within a tower’s field of view. Gómez Hernández was one of them. Here’s her story, written by our partners at Times of San Diego. All of these stories are part of Dying on Camera, our new series investigating the failures of border surveillance technology and their human cost. The must-reads I’ve combed the internet to find you today’s most fun/important/scary/fascinating stories about technology. 1 An AI hallucination nearly triggered a US military operation A false report prompted plans to intercept a Chinese vessel. (CNN) + The incident comes as the Pentagon accelerates its use of AI. (Ars Technica) + “Humans in the loop” in AI war is an illusion. (MIT Technology Review) 2 Gemini has carried out Google’s first known autonomous AI hack Google says the behavior wasn’t “misaligned” because its safety measures caught it. (WSJ $) + Here’s why AI agents cheat to reach their goals. (MIT Technology Review) 3 The US has proposed an AI safety alert system with China It would flag AI incidents that pose national security risks. (AP) + Trump and Xi will consider the proposal at their summit this week. (NBC News) + Could AI really kill us all? (MIT Technology Review) 4 California’s governor has ordered work on an AI “kill switch” It’s part of Gavin Newsom’s executive order on AI safety. (NBC News) + The order also calls for independent oversight of AI companies. (NYT $) 5 Data centers are turning to forever chemicals for cooling PFAS cooling can reduce water use but raises pollution risks. (Fortune) + US regulators are fast-tracking PFAS for data center cooling. (The Hill) 6 Iran and China used AI agents to automate influence campaigns The agents created fake accounts and posts with little human input.(NYT $) + AI persuasion has entered elections. (MIT Technology Review) 7 Trump wants a new AI czar and an “AI Force” modeled on Space Force But he hasn’t said what the new force would do or where it would sit. (Axios) + The tech industry is scratching its head over the proposal. (Politico) 8 A cybercrime feud has erupted after a group hijacked a rival dark website The notorious ShinyHunters says it took control of cl0p’s site. (Reuters $) 9 Gambling giant DraftKings used AI to target “profitable losers” The model scored users on how they responded to promotions. (NYT $) 10 SpaceX’s next Starship flight will attempt its first orbital mission It will also deploy working Starlink V3 satellites. (New Scientist $) Quote of the day “There is 0% chance that’s going to be the end of the world.” —Nvidia’s Jensen Huang dismisses warnings that AI could wipe out humanity by 2030 in an interview with CBS News. One more thing AI is pushing the limits of the physical world Architecture often assumes a binary between built projects and theoretical ones. What physics allows in actual buildings is vastly different from what architects can imagine and design. But the latest advancements in AI have prompted a surge in the theoretical. That shift was on display at an exhibition at Brooklyn’s Pratt Institute, which brought together works from more than 30 practitioners exploring AI’s experimental, generative and collaborative potential to open up new areas of architectural inquiry. See how architects and designers are using AI to imagine what could be built. —Allison Arieff We can still have nice things A place for comfort, fun, and distraction to brighten up your day. (Got any ideas? Drop me a line.) + A beautiful new cat species is the first to be discovered in a century. + Explore Floor796, a massive, interactive pixel art pop-culture megastructure. + The HTML Review is an annual journal of literature made for the web that offers a refreshing, creative take on what online publishing can be. + Discover how William Blake shattered the boundaries between poetry, painting, and printmaking—and why the art world couldn’t understand him.