Many frontier labs keep their risky research locked away. Trillium Labs wants to show off its work when it comes to self-improvement and model behavior. Some big AI companies seem to think that the best way to keep models from becoming too chaotic or too mischievous is to keep them locked inside of labs. If only a chosen few can access them, the thinking goes, they can do less damage in the real world while researchers try to understand what they're capable of. Nathan Lambert and Tom Zick, two industry scientists, believe the opposite. The pair founded a nonprofit, Trillium Labs, that will work on various areas of AI research--including potentially problematic areas like recursive self-improvement (RSI) and agents--in a more transparent way.

Trump's Crazy AI Rebrand Was a Loyalty Test for Tech Execs--and It Worked But they meekly went along when Trump cancelled the term. Seventy years ago last summer, a group of scientists met at Dartmouth College to discuss a wild idea--that machines might one day be capable of human-like thought processes. One of those scientists, John McCarthy, had suggested a name for the new field: artificial intelligence . When you search Google Scholar for "artificial intelligence," approximately 7,420,000 results appear. The term is utterly ubiquitous. Even for people who despise AI, the least of their worries is what we call it.

Amazon Says It's No Longer Using NDAs for Data Centers The company announced it will stop using secret agreements with county officials and acknowledges community backlash could lead to data center "moratoriums" across the country. Amazon has made a series of new public commitments to communities where it's building data centers, including making an official announcement that it no longer uses nondisclosure agreements in its arrangements with county officials. In a blog post published Friday, CEO Matt Garman detailed the steps the company says it's already taking around data centers. In addition to stopping the use of NDAs, Garman says $1 billion will be invested over the next five years in communities that house the company's data centers. The money will go toward free community college programs, workforce training, and energy efficiency upgrades for schools, homes, and community buildings.

Whatever AI Safety Is, It's Not This Asking AI companies to self-regulate is a great way to pretend like you've accomplished something. Don't worry, we'll make it quick! In 1966, nearly 51,000 people died on US highways. The overwhelming evidence suggested that many lives could have been saved with modern seat belts, which by then had been around for several years. The automotive industry responded by rallying together, and, in a triumph of self-regulation, voluntarily put the safety feature in every new vehicle that rolled off the line.

Robert A. Pape, a leading authority on political violence, wants Americans to understand that the threat is coming not from the fringes, but from the mainstream. The attack on the Capitol on January 6, 2021, was not conducted mainly by extremists or fringe militia groups . It was, according to the data, the work of normal Americans. That is the conclusion Robert A. Pape lays out in his new book, citing data about those who were arrested and charged following the attack. While Proud Boys and Oath Keepers gained most of the headlines, data collected by Pape as part of his work as founding director of the Chicago Project on Security and Threats (CPOST) found that the vast majority (90 percent) of those charged with crimes arising from the attack held no affiliation with militias or militant organizations. Pape's analysis found that 22 percent were business owners, and almost 30 percent held a college degree.

OpenAI's Dots and Meta's Muse are vying to be your AI agent of choice. I've tried both--and I suspect you will too. Yesterday, I spent the morning at OpenAI's annual DevDay, where over 2,500 nerds discussed building software and marveled at the latest artificial intelligence models. The biggest reveal of the day was Dots, the company's new, always-on agents that proactively help complete tasks and make autonomous decisions as they scour the web. After watching the sometimes messy demos on stage and hearing CEO Sam Altman talk about his experiences using the agents, I was curious what it felt like to boss my own Dot around.

These cute agents are designed to connect to your apps and tackle multistep tasks. Dots are OpenAI's new always-on AI agents, announced at the company's DevDay 2026 event on Tuesday in San Francisco. These agents complete tasks on a user's behalf and are depicted as cute blobs you can personalize and assign multi-step projects to complete. A core selling point is their "always-on" design--unlike a single-turn chatbot experience, these agents constantly crawl the web and attempt to complete whatever task you assign to them. Dots also pull context from connected apps and learn more about your preferences over time.

Democracy blindsides Silicon Valley's power players looking to transform the world with AI'The push against datacenters speaks to the frustration of people who don't feel they are making the choices in their lives' The Stanford ethicist Rob Reich was once invited to a dinner organized by a Silicon Valley mogul to discuss what a state designed to maximize science and tech powered by commercial models might look like. The gathered technologists were drawn to the idea of building a small nation-state dedicated to the endeavor, Reich recalls in " System Error: Where Big Tech Went Wrong and How We Can Reboot it". When Reich asked whether this state would be a democracy, he got a cool no. "Democracy is too slow, and it holds science back," he was told. "To optimize for science, we need a beneficent technocrat in charge."

Once seen as nerdy but harmless innovators, men in tech are now widely regarded as red-pilled Elon Musk wannabes. It's made getting laid a whole lot harder. Engaged in a flirtatious conversation at a bar in Manhattan, 23-year-old Gary stood with a Guinness in hand as he explained to a woman why he took an engineering job at Palantir, the data analytics company cofounded by Alex Karp and Peter Thiel. Raking in a $250,000 salary would allow him to help pay off his parents' mounting medical bills, he explained. Gary wasn't met with sympathy. Medical debt, the woman told him, was "not an excuse to work somewhere like ."

The company is bringing its Private Processing encryption service to its much-maligned smart glasses. Meta knows you probably have some privacy concerns about its smart glasses . At its annual Meta Connect event today in Menlo Park, California, the company announced plans to incorporate its siloed AI processing scheme known as Private Processing into its Ray-Ban Meta smart glasses. The feature, designed for WhatsApp last year, allows the glasses' AI assistant to get to know you and fulfill your requests without Meta accessing any of your data. "Most of the time, glasses are helping you see well, protecting your eyes and complementing your look, and that's it," Meta wrote in its press release, perhaps somewhat underselling the monumental historic significance of vision-correcting lenses as a technology.