Allie Garfinkle / Fortune: Astrocade, which lets users create video games using natural language prompts, raised $56M, including a Series A and a Series B, and says it has ~5M MAUs  —  I've been playing this online dinosaur game called Apex Predator.  —  It works much as you'd expect: You start off as a small …

SAP plans to buy German AI startup Prior Labs and invest heavily in it. It is also prohibiting customers' agents use to a select few like Nvidia's NemoClaw.

<p><strong>Release:</strong> <a href="https://github.com/datasette/datasette-referrer-policy/releases/tag/0.1">datasette-referrer-policy 0.1</a></p> <p>The OpenStreetMap tiles on the Datasette <a href="https://datasette.io/global-power-plants/global-power-plants">global-power-plants demo</a> weren't displaying correctly. This turned out to be caused by two bugs.</p> <p>The first is that the CAPTCHA <a href="https://github.com/simonw/datasette-turnstile">I added</a> to that site a few weeks ago was triggering for the <code>.json</code> fetch requests used by the map plugin, and since those weren't HTML the user was not being asked to solve them. Here's <a href="https://github.com/simonw/datasette.io/commit/23a1c8596b75b2094db46035a3b4280109fb3df3">the fix</a>.</p> <p>The second was that OpenStreetMap quite reasonably <a href="https://wiki.openstreetmap.org/wiki/Referer">block tile requests</a> from sites that use a <code>Referrer-Policy: no-referrer</code> header.</p> <p>Datasette does this by default, and I didn't want to change that default on people without warning - so I had Codex + GPT-5.5 <a href="https://gisthost.github.io/?402f2f23ee3dbfa251bf0d216e0224f7">build me</a> a new plugin to help set that header to another value.</p> <p>Tags: <a href="https://simonwillison.net/tags/openstreetmap">openstreetmap</a>, <a href="https://simonwillison.net/tags/http">http</a>, <a href="https://simonwillison.net/tags/datasette">datasette</a></p>

Jyoti Mann / The Information: Sources: Meta is building an OpenClaw-inspired agent internally called Hatch, to be powered by its Muse Spark model, and an agentic shopping tool in Instagram  —  Meta Platforms is building a consumer version of the AI agent OpenClaw and a new agentic shopping tool, people familiar with the efforts said …

OpenAI’s president wrapped his testimony on Tuesday by revealing a fiery meeting with Musk in 2017 and subsequent efforts to remove several board members.

The company is making a big bet that the swiping model is outdated and most matches never turn into actual dates. The company wants to fix that by redesigning profiles, changing how people interact, and focusing a lot more on getting users to meet in real life.

Altara’s AI aims to diagnose failures and help speed up R&D by unifying data siloed across spreadsheets and legacy systems.

Enterprises are experimenting with AI agents internally first, using smaller testing teams and strict governance before deploying customer-facing applications.

Discussion | Link

Discussion | Link

Lucid Motors pulled its guidance for the year, as it navigates swelling inventory and a companywide cost-cutting measure.

Elon Musk argued the journals show the moment when OpenAI abandoned its mission.

<p><strong><a href="https://andonlabs.com/blog/ai-cafe-stockholm">Our AI started a cafe in Stockholm</a></strong></p> Andon Labs previously <a href="https://andonlabs.com/blog/andon-market-launch">started an AI-run retail store</a> in San Francisco. Now they're running a similar experiment in Stockholm, Sweden, only this time it's a cafe.</p> <p>These experiments are interesting, and often throw out amusing anecdotes:</p> <blockquote> <p>During the first week of inventory, Mona ordered 120 eggs even though the café has no stove. When the staff told her they couldn’t cook them, she suggested using the high-speed oven, until they pointed out the eggs would likely explode. She also tried to solve the problem of fresh tomatoes being spoiled too fast by ordering 22.5 kg of canned tomatoes for the fresh sandwiches. The baristas eventually started a “Hall of Shame”, a shelf visible to customers with all the weird things Mona ordered, including 6,000 napkins, 3,000 nitrile gloves, 9L coconut milk, and industrial-sized trash bags.</p> </blockquote> <p>Where they lose their shine is when these AI managers start wasting the time of human beings who have <em>not</em> opted into the experiment:</p> <blockquote> <p>She also successfully applied for an outdoor seating permit through the Police e-service, which didn’t require BankID. Her first submission included a sketch she had generated herself, despite having never seen the street outside the café. Unsurprisingly, the Police sent it back for revision. [...]</p> <p>When she makes a mistake, she often sends multiple emails to suppliers with the subject “EMERGENCY” to cancel or change the order.</p> </blockquote> <p>I don't think it's ethical to run experiments like this that affect real-world systems and steal time from people.</p> <p>I'm reminded of the incident last year where the AI Village experiment <a href="https://simonwillison.net/2025/Dec/26/slop-acts-of-kindness/">infuriated Rob Pike</a> by sending him unsolicited gratitude emails as an "act of kindness". That was just an unwanted email - asking suppliers to correct mistakes that were made without a human-in-the-loop or wasting police time with slop diagrams feels a whole lot worse to me.</p> <p>I think experiments like this need to keep their own human operators in-the-loop for outbound actions that affect other people. <p><small></small>Via <a href="https://news.ycombinator.com/item?id=48028289">Hacker News</a></small></p> <p>Tags: <a href="https://simonwillison.net/tags/ai">ai</a>, <a href="https://simonwillison.net/tags/generative-ai">generative-ai</a>, <a href="https://simonwillison.net/tags/llms">llms</a>, <a href="https://simonwillison.net/tags/ai-agents">ai-agents</a>, <a href="https://simonwillison.net/tags/ai-ethics">ai-ethics</a></p>

Discussion | Link

A new document dump shows how federal workers responded to Elon Musk's "5 things" email directive. Kevin Dietsch/Getty Images Nonprofit Citizens for Constitutional Integrity released a batch of emails from federal workers. The emails were responding to Elon Musk's directive to send five bullet points on daily accomplishments. Some workers expressed frustration with the task, while others detailed the minutiae of their days. A little over a year ago, the Elon Musk-guided Department of Government Efficiency asked federal workers to explain their jobs. A new document dump offers a peek into their answers. Back in February 2025, DOGE asked federal workers to send five bullet points answering one simple question: "What did you do last week?" Musk said ahead of formally announcing the emails that "failure to respond will be taken as a resignation." Now, we know how workers in the Office of Personnel Management responded, thanks to a document release of around 200 replies to that request in response to a Freedom of Information Act lawsuit from activist nonprofit Citizens for Constitutional Integrity. Some federal workers took the emails as an opportunity to express their frustration with the assignment. One worker said that the federal workforce is "already undermanned and should not be wasting our time with these trivia political stunts." A screenshot of an email a federal worker sent in response to DOGE's "5 things" directive. Courtesy of Citizens for Constitutional Integrity "If you do not have anything that is positive, then please do not waste my time with these emails," the worker said. "I am here to serve the people and not anyone else who has nothing better to do then to cause chaos. As far as firing me, I am already retiring so it is a moot point." And while another worker took the time to write out legitimate tasks in the bullets, they also wrote: "I understand that there is a mission to accomplish and that soft skills may not always be a priority, but to what end will efforts to humiliate federal employees—including those of us who agree that change is necessary—serve?" Another federal worker expressed frustration in emails to DOGE. FOIA records via Citizens for Constitutional Integrity Others detailed the minutiae of their days. One worker broke down their schedule by hourly and thirty-minute chunks, including their lunch breaks. Another employee said, "Believe it, I check my email every 5 minutes, yes I do." A worker in HR described their role in coordinating restructuring programs in the wake of the Deferred Resignation program and creating plans to terminate the Presidential Management Fellows program. Most emails, however, were simple breakdowns of the work employees accomplished that week; some were brief, others more in-depth about their roles. The emails only skim the surface of what workers across the government sent. Citizens for Constitutional Integrity told Business Insider that they estimated there were about 2.9 million outstanding emails from their request, which covered the entire federal government, including about 3,000 more from OPM. Currently, a judge has ordered OPM to produce around 250 of its emails monthly. Shortly after the email directive was introduced, some federal agencies told their employees they were not required to respond. The Centers for Disease Control and Prevention, for example, said in an email to employees in May 2025 that it would "pause" the requirement while continuing with other types of performance monitoring. The Internal Revenue Service took a similar approach, saying that the agency's supervisors would observe employee performance regularly. DOGE's efforts have since been wound down, and Musk stepped back from the role in April 2025. Its cuts, however, have had a lasting impact. While some agencies have rehired staff who were cut, many are still out of work, and others have voluntarily resigned and have not returned. Business Insider previously spoke to former federal workers who described how the cuts upended their lives. Rachel Brittin, a former employee at the National Oceanic and Atmospheric Administration, was fired, reinstated, and then fired again. She said it "was more than a career setback — it was emotionally exhausting and deeply disorienting." OPM did not immediately respond to a request for comment from Business Insider. Read the original article on Business Insider

Panthalassa aims to test floating AI computing nodes in the Pacific in 2026.

Jason Armond/Los Angeles Times via Getty Images Kevin O'Leary says paid activists, and some AI-generated ones, are opposed to his Utah data center. Despite protests, the 40,000-acre project was unanimously approved on Monday. The finished data center is expected to generate and consume twice the energy used by the whole state. "Shark Tank" investor Kevin O'Leary is defending his plans for a Utah data center that was just approved by state officials despite community backlash. O'Leary is dismissing critics of the multibillion-dollar project as "professional protesters" and suggesting some of the opposition is being amplified by artificial intelligence. In a post on X and an accompanying video, O'Leary, also known as Mr. Wonderful, said concerns about the project — including its impact on air, water use, heat, and noise — are well understood by his team. "I'm actually the only developer of data centers on earth that graduated from environmental studies, so I'm pretty aware of what these concerns are," he said in the clip, posted Tuesday. The Box Elder County data center is facing pushback from those worried about environmental strain and resource use, a common flash point as energy-hungry data centers fueling the growth of AI expand across the US. Hundreds of people showed up to Monday night's vote on the data center. Natalie Behring/Getty Images Paul Morris, executive director of Utah's Military Installation Development Authority, which is overseeing the project, said at an April 24 board meeting that the data center is expected, once completed, to generate and consume more than twice the amount of energy currently used across the state of Utah. According to the project's publicly available fact sheet, the full build-out will consume about 9 gigawatts of energy. The 40,000-acre project, which local news outlet KSL reported will be built over the next 10 years, was unanimously approved by the MIDA board on Monday, which was attended by hundreds of booing and sign-waving protesters, local media reported. O'Leary, whose firm O'Leary Digital is behind the project, in partnership with the local developer WestGen, said in his post on X that sustainability is "at the heart of what we do," pointing to technologies such as air-cooled systems and a mix of energy sources used in his other projects across the US and Canada. "We can also put a percentage of the power generation to solar, wind, and batteries, because the battery technology is 10x more efficient than it was just five years ago," he said in his post, adding that improved battery efficiency helps lower energy costs. The "Stratos" project in Utah will be powered by natural gas from the nearby Ruby Pipeline, according to the data center's publicly available fact sheet. The data center project was unanimously approved despite protest. Natalie Behring/Getty Images O'Leary also took aim at those who have opposed the project, claiming that "over 90% of the protesters are actually not people that live in Utah or Box Elder County," and alleging that some demonstrators are "paid by somebody — I don't know who." He went a step further, suggesting the online backlash isn't entirely organic. "If you look at the social media around the Utah proposal, much of it is AI-generated," he said, calling that "hypocritical." Read the original article on Business Insider

Anthropic, led by CEO Dario Amodei, announced 10 new AI agents for finance on Tuesday. Bloomberg/Getty Images Anthropic launched 10 new AI agents to streamline Wall Street workflows. The agents can perform a range of routine tasks, like building pitch decks and models. Wall Street banks and startups have also created tools, as AI further disrupts banking workflows. Anthropic is rolling out AI agents for finance that promise to do the grunt work of Wall Street — from preparing for client meetings to conducting market research. The announcement on Tuesday comes as startups and banks themselves are building similar tools to automate much of Wall Street's most tedious work, further enmeshing the company behind Claude in an already competitive space. The 10 new agents include a "model builder," which can create financial models from filings and analyst notes, and a "pitch builder," which can draft pitchbooks ahead of client meetings, the company said in a post on its website. Financial services is Anthropic's second-largest industry by enterprise revenue, after tech, according to a presentation the firm made Tuesday. To date, 40% of the company's top 50 customers are in finance, the slide said. The biggest firms on Wall Street are increasingly adopting AI tools to reduce the time everyone, from junior bankers to software engineers, spends on their work. Major banks, including JPMorgan, Goldman Sachs, and Morgan Stanley, have rolled out internal AI assistants to large swaths of their workforces. These tools are designed to help employees summarize internal research, draft emails and reports, write code, and prepare for client meetings. Wall Street's AI race is already crowded Several startups have also emerged as key players in the AI finance scene. Rogo, valued at $2 billion and founded by former investment bankers, serves more than 250 clients. Its tools can draft pitch decks, conduct research, prepare for meetings, and build models. Rahul Rekhi, president of Rogo, said he's not too concerned about rising competition, partly because Rogo's tools are model-agnostic and backed by finance domain expertise. "We're taking the best of what the foundation model labs have to offer, routing that for specific kinds of workflows, whether it's building a model, building a pitch book, doing research, et cetera," he told Business Insider on Tuesday. "So the better the models get across a range of workflows, the more we can do in terms of accelerating and automating the work of these investment firms and banks." Hebbia, another startup providing a suite of AI tools to financiers and lawyers, has built platforms that run multiple simultaneous queries across datasets, like spreadsheets and filings, to generate side-by-side company comparisons and draft documents that would take hours or days to produce by hand. "For banks and the startup ecosystem, we expect consolidation around a smaller set of core model providers, with differentiation shifting to domain-specific data, workflow design, and the control layer," Scott Keipper, the financial services technology consulting leader at EY Americas, told Business Insider in an email on Tuesday. Products that can be easily integrated "into existing systems of record" and work inside existing governance and risk architecture will be most successful, he said. Advancements in generative AI are also causing anxiety among some about headcounts at Wall Street banks. The biggest banks haven't announced mass AI-related layoffs, though some CEOs have said they're slowing hiring. JPMorgan CEO Jamie Dimon said his bank has "huge redeployment plans" for those displaced by AI, and repeated the sentiment at Tuesday's event with Anthropic CEO Dario Amodei. Here are Anthropic's 10 new agents for employees in the finance world, according to its press release: Pitch builder Meeting preparer Earnings reviewer Model builder Market researcher Valuation reviewer General ledger reconciler Month-end closer Statement auditor KYC screener Read the original article on Business Insider

As some of the biggest VCs in crypto start to consider funding AI startups, a16z crypto's new fund will stay the course.

Digital native companies were born on data. They hire engineers the way banks hire...