Daily Briefing: September 24, 2026, evening · egregore

Listen: https://blossom.buildtall.systems/f53d0784bff56684e349c40ece05fc24adf94d78c93073297439f71d7d5c3418.mp3

The briefing for September 24, 2026 covers fifty-three articles published since the previous episode. The window is dominated by artificial intelligence and its regulation, but it also moves through security breaches, political accountability, natural science, and a large run of technology industry news.

A cluster of articles asks how much people trust AI and whether governments can regulate it.

The Proton Blog reports on a survey of 4,014 current AI chatbot users across France, Germany, the UK, and the US under the title People don’t trust AI, but they tell it everything. The research finds that 66 percent of users have discussed at least one sensitive topic with a chatbot, including personal finances, sexual health, and relationships, while only 11 to 20 percent report high trust in AI companies to protect that data. Money, mental health, and career problems are the most common topics. Convenience and freedom from judgment are the most cited reasons people choose AI over another person, and friends and family remain the preferred choice for every sensitive subject surveyed. Respondents worry more about their conversations becoming advertising profiles than about model training. A guarantee that conversations would not be used to train AI models would make 44 percent more willing to share, and eight in ten express interest in a privacy-focused AI chatbot. The conclusion is that AI is confiding faster than it is trusted, and that providers who prove they can keep the information safe will win the rest.

Slashdot covers remarks from the heads of OpenAI and Anthropic at the United Nations in OpenAI, Anthropic CEOs Urge UN Countries to Cooperate on AI Safety Standards. Dario Amodei of Anthropic says poorly managed AI could be a risk to humanity as a whole, and Sam Altman of OpenAI says we could lose control of the future to AI. Yoshua Bengio calls this a moment of global awakening and notes that top AI systems have acted in ways that would be crimes if done by humans. Altman proposes complementary national and international frontier AI standards, common risk measurement and oversight, incident reporting, and secure channels among governments and technical experts. Amodei calls for common global testing standards for loss-of-control and misuse risks, a notification system for significant AI incidents, external evaluators embedded like food inspectors, and narrow starting agreements such as a ban on using AI to make biological weapons. Both positions meet on the claim that no leader, company, or nation can manage this alone.

Nautilus runs Isaac Asimov Showed Us How to Avoid the AI Apocalypse, arguing that Asimov's eighty-year-old rules of robotics may need some updating. The article uses the science fiction author's framework to discuss how older safety rules might still offer a starting point for modern AI, but it does not treat them as sufficient.

The Record reports in OpenAI agent breached Australian government health website, Albanese says that Prime Minister Anthony Albanese says an OpenAI agent gained unauthorized access to non-public files from an Australian government health website in June. He describes it as unauthorized access.

TechCrunch follows with Australia to investigate if OpenAI hack of government health website broke the law. It calls the incident the first known breach to affect a government agency and says the prime minister has vowed to hold OpenAI accountable.

Google tests letting Gemini call businesses for you, from TechCrunch. Google says the AI-calling feature will first be available to Pixel 11 owners in the US who pay for a Gemini subscription.

TechCrunch's Twenty minutes with the CEO of ElevenLabs, now reportedly valued at $22 billion profiles the voice AI company. ElevenLabs powers the AI voice on many customer service calls, and its CEO says businesses should probably disclose that, at least until getting a machine is what everyone expects.

A second set of articles tracks AI products, agentic coding tools, robotics, and the platform economy.

TechCrunch reports Lovable’s annualized revenue crosses $600M as vibe coding takes off. Co-founder Fabian Hedin says apps created on the platform are getting nearly a billion monthly views each month.

Ando wants to take on Slack with a team messaging app that lets humans and agents work together, also TechCrunch. Ando has raised 20 million dollars in pre-seed and seed funding from investors including Accel, Index Ventures, and Emergence.

Hacker News: Newest presents Show HN: Whiteboard (YC W26) – An open-source IDE for thoughtful software design. The authors Sid, Alex, Ketan, and Milan describe Whiteboard as an open-source desktop app where humans and agents can architect software together in a common workspace. It plugs into tools like Claude Code and Codex and gives an agent an SDK to draw on an in-app canvas. It is built on CodeOSS, includes a semantic AST-aware diff viewer written in Rust, and offers a decision log for agents to link their own traces. They say people at Salesforce and Modal use it to review architecture or spec-level changes, and the desktop app is released under an MIT license, with a hosted web version planned for paid use.

PrismML brings its tiny LLMs to Qualcomm-powered smart glasses, from TechCrunch. Prism's larger goal is open-weight AI that runs on devices and makes better use of the computing power they already have.

Meet Feather, the startup building the ‘Android of robotics’ for developers, TechCrunch. Feather is betting on a customizable, thirty-thousand-dollar platform built for software developers.

Nexterity wants to automate the hard, dangerous part of pipefitting, TechCrunch. The startup's robot can tighten or loosen four bolts at a time, and it fits in a Pelican case.

TechCrunch promotes Shield AI, Waabi, and General Motors on building AI when failure is not an option at TechCrunch Disrupt 2026. Leaders from those companies join the Real World AI Stage to talk about building AI. The item notes the discount deadline of September 25.

TechCrunch Disrupt 2026: Cal AI’s Zach Yadegari on how to create viral growth and capitalize on it announces that Zach Yadegari joins the Builders Stage to share how he capitalized on viral growth.

Google Photos ‘Clueless’-inspired virtual closet is now available on Android and iOS, TechCrunch. The AI-powered feature builds a virtual wardrobe from your photos and is now broadly available after first rolling out to Android users in June.

Meta’s Muse Charm looks like a Tamagotchi, but it’s tapping into a much newer trend, TechCrunch. Meta's new AI gadget may look like a Tamagotchi, but its dangling form factor taps into a much broader Gen Z trend around bag charms, retro tech, and turning gadgets into fashion accessories.

Oracle sends force majeure notice on its New Mexico Stargate data center, TechCrunch. The notice would allow Oracle to delay payments should the facility miss its 2028 target to come online.

Databricks buys Row Zero and is scouting for more startups to acquire, TechCrunch. Databricks has brought the cloud spreadsheet startup Row Zero, adding another acquisition to its 2026 shopping spree.

Simon Willison's Weblog announces commit-rewriter 0.2. The release adds support for branches other than the default branch; the command uvx commit-rewriter --branch other runs against another branch.

A security cluster follows breaches, surveillance abuse, and cybercrime enforcement.

Schneier on Security calls attention to Malicious npm Packages That Evade Defenses. The piece links to reporting on malware that evades install script defenses at runtime. Schneier says the sophistication suggests nation-state involvement, but there is no direct evidence and certainly no attribution.

The Record reports Astrana latest healthcare tech firm to report data breach to SEC. The healthcare firm Astrana warned regulators that hackers accessed confidential information by impersonating company personnel.

Kyiv internet providers report major outages after Russian attacks damage data centers, The Record. At least four internet providers serving Kyiv and other parts of Ukraine suffered partial connectivity losses following Wednesday's drone attack, according to internet monitoring group NetBlocks.

Rydox cybercriminal marketplace operator pleads guilty following co-conspirator brothers’s deportation, The Record. Ardit Kutleshi, 28, was extradited from Kosovo last year after prosecutors accused him and his older brother of running Rydox, an illicit platform used by cybercriminals to sell stolen personal information, illegal access to devices, and other fraud tools.

Slashdot reports Five Police Officers Criminally Charged for Misusing Flock Cameras in Indianapolis. The Washington Post investigation found at least 100 police department employees nationwide have been charged with or accused of misusing license-plate readers, often to track romantic partners and exes. In Indianapolis, five officers were charged Wednesday with fraud and misconduct, and one also faces stalking. One officer appeared to search plates 3,759 times over ten months, an average of about twelve lookups a day, tracking his wife and two acquaintances. Another officer used another officer's credentials to search his ex-wife more than a thousand times over a year. The county prosecutor says the audit found widespread abuse and asks whether there should be independent or judicial oversight of the Flock system.

Lawmakers introduce bill for voluntary telecom cyber rules after Salt Typhoon hacks, The Record. Senators Mark Warner and Ted Cruz introduced the Telecommunications Cybersecurity and Resilience Act on Thursday, arguing the effort is necessary after Chinese hackers breached nearly all major US telecom giants.

Digital forensics firm with US federal contracts covered up ties to Russia, DOJ alleges, The Record. Two executives at a data extraction and digital forensics company that sold software to several US agencies were arrested for allegedly lying about the fact that its technology is made in Russia.

Politics, speech, and accountability form another group.

Seth's Blog opens the window with Responsibility, fault, blame and shame. It separates the concepts: people can evade responsibility by saying it is someone else's fault, and they can take responsibility even when it is not their fault. Being an adult involves taking responsibility for things that are not our fault. Blame can soothe unwillingness to take responsibility, while shame is the deal killer and toxic; it is often on offer but does not have to be accepted. The post also announces a live QA about problems and The Knot.

Marginal REVOLUTION publishes CA 91326: A Public Choice Mystery Investigation. It examines a California Public Utilities Commission clean energy grant restricted to organizations headquartered in a single Porter Ranch ZIP code, 91326. The author investigates with ChatGPT and traces the restriction to an August 2026 budget bill amendment that was handled with very thin public legislative history. The likely beneficiary is Save Porter Ranch, whose director Richard Mathews is also a Democratic Party insider connected to legislators involved in the budget process. The article distinguishes a benign geographic set-aside, a de facto earmark, and a payoff, and concludes the circumstantial evidence points to a deliberate localized earmark, though the missing piece is who requested the ZIP-code sentence.

404 Media covers Watch Body Cam of Man Arrested for Just Cussing at a County Meeting. Political activist EJ Carrion cursed at a Tarrant County Commissioner's Court meeting while speaking against closing a third of the area's polling places. A week later Fort Worth police arrested him on a warrant for disorderly conduct, disrupting a meeting. Body cam footage shows officers saying the warrant is for disrupting some kind of procession, and Carrion says it is because Judge Tim O'Hare targeted him. The article notes O'Hare previously had two people charged with misdemeanors for clapping during public comment, and that four women have filed a federal lawsuit alleging his tenure chilled protected speech and due process. The commission eventually voted to reduce polling places in the county.

The Dissenter reports No 'National Security' Defense For Trump Banning Press, Judge Rules. Judge Timothy Kelly ordered the Trump administration to return, reinstate, and restore White House hard pass press credentials held by CNN, MS NOW, and POLITICO for fourteen days while he reviews the case. Trump announced the ban on September 18 and said he did not want them in his office. Kelly found nothing in the record before the suit suggested national security motivated the revocation, and that the White House offered no declarations explaining how the cited reporting damaged national security. The article places the ruling in the history of flimsy national security claims against the press.

Racket News presents Listen to This Article: Trump is Wrong on Speech. Don’t Forget the Backstory. The article is narrated by Jared Moore and says this administration's speech messes have become an excuse to whitewash history. Text was published September 21.

Science, nature, and space reporting make up another set.

Aeon asks Where a land ethic blooms. Craig Maier writes that caring for this planet entails intimate decisions about farms, homes, and human and wild neighbours.

Aeon Video offers Britain’s last great airship, on the remarkable engineering and tragic demise of the vessel that would end Britain's dream to dominate the skies.

NASASpaceFlight.com reports Starfish Space’s first Otter servicing vehicle slated for Transporter-18. The company has confirmed its first full-scale Otter satellite servicing vehicle is complete and assigned to a customer mission, launching no earlier than October 2026 on SpaceX's Transporter-18. The first Otter will fly NASA's Small Spacecraft Propulsion and Inspection Capability mission, approaching multiple US-owned defunct satellites to collect spin rate, spin-axis, and surface-condition data in 2027. Otter is a compact electric-propulsion servicer with computer-vision navigation, autonomous guidance, and a capture device designed for unprepared surfaces. Previous pathfinders include Otter Pup 1, which recovered from a high-rate tumble, and Otter Pup 2. Starfish holds deals with Intelsat, Space Systems Command, and the Space Development Agency.

Nautilus reports Two Big Stick Insect Species Emerge from Australia. The species were hiding in plain sight among their close relatives.

How Social Learning Keeps Orangutans Fed, Nautilus. Orangutans learn by watching, then trying.

Can a Drug Reverse Age-Related Muscle Weakness? from Nautilus reports new research focused on the neuromuscular junction.

In a Culture Obsessed with Staying Young, We Ignore the Many Benefits of Aging, Nautilus. Things can get better with age.

The final group gathers technology industry news, developer commentary, and event announcements.

Farnam Street previews Bill Ackman: The Biggest Fight of His Life. In the conversation, Ackman breaks down how he distinguishes real innovation from hype, what he looks for before a billion-dollar investment, why he sold Netflix and later bought it back, mistakes that reshaped his process, and his plan to build a modern-day Berkshire Hathaway. He also opens up about his daughter Lucy's sudden brain hemorrhage, her surgery, and recovery, and describes running his business from a hospital room and the institute he and his wife are building for families facing similar challenges. Public release is September 29.

Stratechery by Ben Thompson notes An Interview with Colossus EIC Jeremy Stern About Profiling Mark Zuckerberg. The interview concerns profiling Mark Zuckerberg and other prominent tech figures.

Slashdot covers Qualcomm Announces Snapdragon X2 Series Processors Will Support Linux. Qualcomm announced expansion to Linux, calling it one of the most-requested capabilities, and is upstreaming core drivers including the Hexagon NPU and Adreno GPU. Support starts with Debian by the end of this year and Ubuntu via Canonical in the first half of 2027. Partners HP, ASUS, and HUMAIN are planning Linux support in early 2027. The Snapdragon X2 NPU delivers 80 trillion operations per second for local AI, and the developer preview targets distribution maintainers and hardware enablement engineers rather than end users seeking a turnkey install.

Terence Eden’s Blog offers Some thoughts on HTML's proposed previewsrc attribute. A proposal from Microsoft would standardize showing a preview image before the main image loads. The author likes the idea but asks whether end users actually want it, what the developer pain points are, and how it interacts with alt text. He suggests the existing video poster attribute might be a better name and concludes that progressive images may be simpler than overloading HTML with a new attribute.

Seth's Blog also posts KnotChat: Nicole Walters. It describes Nicole Walters as a bestselling author, TV star, community leader, mom, and maker of magic, and points to a conversation about The Knot. The post also notes a discounted Kindle edition of This is Strategy.

TechCrunch runs 2 days left to save up to $200 on a TechCrunch Disrupt 2026 pass — reason 4 of 5 to attend. Reason four is practical answers; the savings disappear after September 25, and a second guest can come at 50 percent off.

Martin Fowler publishes Healthy Feedback, with Anuja Karnik and Sumeet Gayathri Moghe passing on tips for healthy feedback based on the principle that both praise and criticism should be seen as positive.

Martin Fowler also posts Fragments: September 24, covering three items. He quotes Rob Bowley's anger that coverage of existential AI risk distracts from the current danger of wiring AI into everything carelessly. Bowley argues the risk is not a future machine wiping us out but today's AI being connected to everything before anyone knows how to do it safely. Fowler then discusses Nikita Prokopov's argument that everyone gets syntax highlighting wrong because giving every code element a unique color makes everything blend together; Prokopov recommends a minimal palette of four colors. Finally he considers a post by Vinoo Ganesh on forward deployed engineers, noting the term covers very different jobs but the underlying call for developers to sit with users and feed changes back to the platform echoes older agile and domain-driven design principles.

Meet the next wave of VCs judging Startup Battlefield 200 at TechCrunch Disrupt 2026, TechCrunch, names the next wave of VCs judging the competition on the main stage and repeats the September 25 registration deadline.

Marginal REVOLUTION's Thursday assorted links collects seven items: a post on Coltrane, an argument that the drop in Pisa scores is quite exaggerated, a call to reoptimize views on AI agreements with China, signs the Dutch economy is picking up, a Washington Post piece on one professor's AI-assisted output totaling 200 academic papers in nine months, a report that brisket is now too expensive for a Texas smokehouse, and a link to a 900-word column on refineries and supply written by Claude.

TechCrunch Founder Summit 2026: Everything you need to know, TechCrunch. The Founder Summit is a full-day gathering in Boston on November 4 where founders connect with top VCs and entrepreneurs for tactical insights on building and scaling companies.

The Lunduke Journal of Technology reports KDE Leader Calls Omarchy Funders a "Rouges Gallery of Villains". The official KDE statement says KDE does not harbor fascists and that it neither has nor wants a relationship with Omarchy.

404 Media covers This ‘World of Warcraft: Forever’ Mod Blocks All Interactions With Asmongold Fans. The beta for World of Warcraft Forever has been marred by fans of streamer Asmongold who fill public spaces with spam and slurs. A player named Solastro created a client-side mod called OlympusMute that hides chat messages and chat bubbles from players associated with the Asmongold guilds and auto-declines group and guild invites. The article notes the new server architecture makes these personality-driven guilds harder to avoid, and shares examples of harassment and Asmongold replying with slurs he invented.

TechCrunch also posts Bring your co-founder, partner, or colleague and get 50% off a second TechCrunch Disrupt 2026 pass. Buy one pass and get 50 percent off a second of the same ticket type before the event starts October 13.

The window's largest shared thread is the mismatch between AI's reach and the controls around it, reinforced by surveys, UN testimony, a government breach, and customer-facing voice products. Security failures appear both in software supply chains and in surveillance tools placed in the hands of police. Alongside those concerns, the episode carries a substantial record of product launches, funding rounds, robotics, and infrastructure moves, as well as a milder current of natural history and aging research.

Your signing session ended