SCROLL NEWS / DISCOVERY
Search headlines
Find the stories shaping the conversation.
Results for " model " 3,152 found 🔀 AI-powered Shuffle
Early DLSS 5 Testing Suggests the RTX 5090's Single Power Connector Might Be the Bottleneck
Community testing of leaked DLSS 5 builds suggests the tech may already be hitting a wall on NVIDIA's own flagship GPU. Tom's Hardware ran the neural rendering model through Cyberpunk 2077, Hogwarts Legacy, and Control on an RTX 5090 Founders Edition, as well as an MSI RTX 5090 Lightning Z....
Apple's New Mac Mini Already Discounted - See the Pre-Order Deals
Amazon has introduced the first cash discounts on Apple's brand new Mac mini, with discounts including two of the M6 models as well as the M5 Pro model. All of these deals are pre-order discounts on the 2026 Mac mini, which officially launches on September 22. Note: MacRumors is an affiliate partner with Amazon. When you click a link and make a purchase, we may receive a small payment, which helps us keep the site running.
Heliophysics Education Community meeting presentation
Presentation | Grades 6-12+ | In this presentation Dr. Samuel Schonfeld of the Air Force Research Laboratory (AFRL) discusses recent modeling efforts and results in heliophysics.
El Niño “supercharged”: what Galápagos corals just revealed stuns scientists - Futura-Sciences
El Niño’s Comeback: Stronger Than Ever? Right now, an unusually intense El Niño is developing in the Pacific Ocean. Recent models from the European Centre for Medium-Range Weather Forecasts (ECMWF) predict that the 2026 El Niño could reach unprecedented strength for autumn and winter. And the effects won’t be limited...
Acer Swift Blade 14 weighs just 799 grams with Intel Core Series 3 processor
Acer has announced a new 799-gram Swift laptop with Intel Core Series 3 hardware. The 14-inch model includes a 50 Wh battery and up to 1TB of PCIe Gen4 storage.
Leaked iPhone 18 Pro Parts Point to Three Color Options Excluding Black
Longtime leaker Sonny Dickson has shared an image of differently colored SIM card trays that he claims are for Apple's upcoming iPhone 18 Pro models, suggesting Apple's flagship lineup will again be available in a trio of colors: Silver, Dark Cherry, and Sky Blue. Apple currently offers the iPhone 17 Pro in Cosmic Orange, Silver, and Deep Blue.
America first, global health last?
The US is reshaping global health aid through bilateral deals and stricter conditions – raising concerns over funding gaps, sovereignty and whether a transactional model can replace decades of multilateral cooperation
August Monthly Review: ChatGPT In Epic
Death, taxes, another frontier lab healthcare launch inspiring truly insane LinkedIn takes. As much as I want to stay away, I repeatedly am the crewman unplugging his ears whenever we pass these sirens. So it’s not the first and probably not the last time we dial up the typewriter rather than tie ourselves to the mast. The full blow-by-blow timeline: Health API Guy is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber. Subscribe The End of the Standalone PHR (Jan 8): ChatGPT Health arrives for consumers, at the time powered by b.well. I was bullish on the approach in that the best way for the PHR to finally solve distribution was to meet consumers were they already are (i.e. When Horizontal Meets Healthcare (Jan 9): OpenAI for Healthcare puts out the shingle for enterprises, with SharePoint ingestion standing in for the EHR integration the product actually needed. Another One: Anthropic’s Healthcare Debut (Jan 16): Claude's version, on HealthEx rails instead of b.well, with Agent Skills for FHIR development as the one true vertical-specific investment. ChatGPT for Clinicians: The Trap Sprung (Apr 24): OpenAI filled in the missing GTM middle with a free product for verified clinicians, effectively peeling the enterprise wrapper off ChatGPT for Healthcare and taking the PLG fight directly to OpenEvidence. OpenAI’s Second Attempt at Health (Jul 31): a relaunch of their patient specific app, now branded Health in ChatGPT, where the privacy partition came out, the connectors thinned, and b.well disappeared in favor of first-party FHIR work against a smaller network Beyond a Shingle So what was announced now? Today, we’re introducing a new electronic health record integration that brings authorized patient context from Epic into ChatGPT for Healthcare, along with the Healthcare Public Data plugin for direct, structured access to official healthcare datasets like PubMed, DailyMed, and CMS Coverage. Together, these capabilities bring ChatGPT closer to the systems and sources healthcare teams trust, while supporting the controls and compliance healthcare work requires. The core portfolio is unchanged from prior announcements we delineated and discussed (aside from the relaunch of patient-facing). Since trade publications still seem to be confusing them by using the wrong names: Enterprise sales motion: ChatGPT for Healthcare Protecting their core product / PLG: ChatGPT for Clinicians Consumer/patient-facing product: Health in ChatGPT So really, this is a feature release (specifically for ChatGPT for Healthcare when it comes to EHR integration). What’s cool about it, though, is that we are moving beyond just putting out a shingle. Everything in the January enterprise launch was horizontalizable: Horizontal tech companies thus generally start verticalization with their “putting out a shingle phase”. Telling the world you’re open for business as a horizontal tech company reliably attracts early inbound interest from buyers who are already trying to force-fit horizontal tools into domain-specific workflows and are eager for any signal that the vendor intends to support their use case more directly. Nothing was really vertical specific, which is exactly what you'd expect from the ultimate hypergrowth horizontal company dipping their toes into specific industries. But eventually you have to stop changing the sign on the door and start changing the product. Can you guess where the horizontal product was going to run out of road? This is the core dilemma writ large: verticalization requires differentiation, while scale economics of consumer and horizontal push toward unification. I’ve buried the lede so deep here it will be painful to some readers, but what should be shocking to no one is that my perspective is that EHR integration is the only path that meaningfully resolves this tension for ChatGPT for Healthcare. Integration with Microsoft Sharepoint is fine, but it is categorically insufficient for any product that hopes to influence clinical decision-making. SMART Money Well here we are, eight months later! Real vertical investment has begun, which surprised a few people. One friend in the industry messaged me: “Absolutely no way Epic gave ChatGPT API access.” Looking forensically at the announcements, the integration is notably read-only across appointment notes, laboratory results, medications, and specialist documentation. So while I’m not a betting man, I'd wager the house this is a SMART on FHIR launch: That sounds a lot like USCDI! The recommended path for the workflow they outline would be SMART on FHIR Their Health AI lead’s post is evocative of both standalone and EHR launch SMART, which another termed “ChartGPT” and “EHR Plugin” The implementation burden for SMART on FHIR is the lowest of the available integrative paths This is the team who literally did patient-facing SMART on FHIR with Epic last month Most importantly, if you’re a horizontal company, you generally still haven’t taken the full plunge and thus (overly) value reusability. SMART on FHIR is the logical compromise: healthcare-specific enough to matter, but standardized enough to reuse. Nobody Asked Judy Another friend asked “I thought they had to go to each health system, or did they go direct to Epic and now health systems opt in?” Given the trust paradigms in healthcare, this functionally cannot be Epic hoovering up all the data for OpenAI and shoving ChatGPT in their customers’ faces (or they would revolt). However, I do believe the aforementioned terrible coverage by TechCrunch and other outlets is probably responsible for that question, given the 325 million reference. When using FHIR as a business associate to an Epic customer, there is no EHR gatekeeping to speak of, as that isn’t how fhir.epic.com works: You register an app You pick your APIs You test against the sandbox APIs You list as “Ready for Production” Hospitals pick your app You test and go live with them Bluntly, nobody had to say yes or no in Verona (nor were they given the chance). Vendor Services (their next developer tier up that I doubt OpenAI is using quite yet in this initial release) certainly has contractual paperwork that some resent, but even there, the twin pressures of mounting antitrust and information blocking make outright gatekeeping increasingly fraught. My friend’s sentiment is, to me, representative of a broader industry neurosis, perhaps a sort of scar tissue of prior eras: people assume the gate is still there and never actually try the door. There are certainly other ways Epic (and any EHR) can put its thumb on the scale, but they are not so dumb as to stand in front of a federally mandated API and play bouncer at this exact moment in time. So I think they can and probably should go deeper, as this release is at best parity and at worst behind vertical-specific competition. OpenEvidence did a basic SMART launch with Sutter Health in February, allowing for more convenient evidence search by providers. They then add patient-context (the equivalent SMART on FHIR flow to what we see here) when Cedars-Sinai went enterprise-wide in May. UpToDate has four distinct applications across deeper workflows like patient engagement listed in Epic Showroom and announced a partnership with Epic to power Art at UGM Abridge and other ambient scribes have invested deeply into the deepest bidirectional clinical copilot workflow with integrations well beyond SMART on FHIR. In that light, the “Who is OpenAI primarily targeting here?” is clear. This release brings them into striking distance of OpenEvidence, but not the others quite yet. They are the only one of the three you can reach without building something truly Epic-specific. When will they go further? Competition is a great motivator to overcome the horizontal demons and start building things that can’t be reused: not across industries, not across EHRs, maybe not past the customer you built them for. That’s the price of actually verticalizing. The question is if and when OpenAI will be willing to pay it. Month in Review Here is the monthly review. As a reminder, this is a regular round-up of the month’s posts and other content to surface things you may have missed across regulation, litigation, interoperability, and beyond. AI assistance is used in these bullet summaries so I can focus on articles. Articles Published: None this month Video Content: The Information Exchange: Standards-based Thruple Edition (Aug 14): Back to school for us too, with Brad reporting in from the CMS Health Tech Ecosystem’s one year anniversary in DC. We get into the January CMS-0057 deadline, the fall rulemaking reading list, and why enrollment still keeps most apps off FHIR. The Information Exchange: Epic Dúnadan Edition (Aug 25): A UGM roundup with Ryan Brickner joining for the first time to check our takes against what the building actually thinks. Rangers, Ergo sitting on top of EHI, and a down-market lineup that needs some cuts. Regulatory: Stacked Deck, Bad Hand (Aug 05): ONC’s website refresh took the entire HIT Policy Committee record with it, so I rebuilt the archive and went looking for the regulatory capture story everyone assumes is buried in there. Stacking the deck and winning the hand turn out to be very different things. The Sixth Generation of Patient Access (Aug 19): A quick catalog of the five generations of patient access we’ve layered on since HIPAA, and the two candidates now competing to be the sixth. One is planned. The other is the market routing around the plan entirely. Court cases: Three Cases Walk Into a Docket (Aug 06): Two surprise settlements cleared the board in a single week, and then Judge Maddox dropped ninety-four pages on Vyne v. Henry Schein. The sleeper is a DMCA holding that makes direct-to-database a considerably riskier business model. Epic v. Health Gorilla: Into the MDL (Aug 10): Nine class actions are headed to Miami, and the Panel signaled it wants to bring the case that spawned them along too. That would leave Epic arguing the requests were obviously fraudulent in one courtroom and unknowable in the other. Amazon v. Perplexity: Agents Are Legalized! (Aug 11): The Ninth Circuit vacated the injunction against Comet, holding that the user is the one accessing the servers rather than the company that built the agent. A real win for agentic access, and a much narrower one than the headline suggests. Veeva v. Epic: Come At Me, Bro (Aug 26): Epic’s response brief wants the dismissal affirmed and, unusually, wants the opinion published as precedent. Buried in it is Epic’s own description of what its non-competes actually prohibit, which current and former employees should read closely. EHRs: The Contract Epic Would Never Sign Today (Aug 04): A 1999 SEC exhibit catches Epic licensing nearly its entire product line, Tapestry included, to the company that became TriZetto. The marketing services menu attached to it is the part that will make you blink. Forecast from Verona (Aug 07): The Epic Almanac makes one argument six different ways: the AI is only as good as the networks behind it. Emmie headlines, Art’s context stack is the most ambitious part, and Penny finally gets an autonomous coding date on the calendar. Works With Epic MyChart (Aug 17): Epic’s first new Showroom category since the death of Workshop certifies hardware instead of software, badge on the box and all. Made for iPhone, but for blood pressure cuffs, and rough news for anyone selling the RPM stack sitting in between. UGM Hot Takes 2026 (Aug 20): Everyone else covered the AI announcements, so I went after Savvy deleting the payment gateway, Rangers as Boost with the timer removed, and a down-market lineup with too many entries. Plus the Health Grid tidbits I cannot help myself on. MyChart Central Grows Up (Aug 24): Device data and Emmie turn Epic’s identity hub into a full consumer platform, which is both a logical answer to ChatGPT and a bit of a mistake. It also happens to be the best scraping target Epic has ever shipped. Industry Analysis: The Wrong Yardstick (Aug 12): Vertical software keeps getting judged against Superhuman and Notion, which is the wrong bar entirely. Your user is comparing your product to a whiteboard and forty phone calls before lunch. Primitives vs. Abstractions (Aug 27): Developers want building blocks, systems of record prefer to hand out business logic, and both sides have a real case. There’s no test from the outside that separates the engineering reason from the competitive one, which is why this keeps ending up in court. Cross-industry Comparisons: The Two Kinds of Platform Power (Aug 18): Attention power and record power are different problems that keep getting handed the same regulatory toolkit. Congress’s latest swing at Big Tech shows both what’s possible and where it falls apart. The Dogs of (Platform) War (Aug 21): A CourtListener alert turned up a property management fight with healthcare’s exact shape and none of the Cures Act. Shell prospects, ghost accounts, a notetaker bot that came back to haunt someone, and a lawyer arguing both sides of the same theory in two states. Fractals All the Way Down (Aug 31): Yardi dominates property management right up until you zoom into one segment, where AppFolio owns it outright and Yardi doesn’t appear at all. Where you draw the line around a B2B software market is about to decide an awful lot of cases. Other News: If Judy Had the Courage (Aug 13): A short eulogy for the era when software was allowed to look insane: WinAMP, bold colors, and everything the grey chatbot era has taken from us. External Media: Portland Monthly Health Tech Meetup: Erin O’Brien channeled my own feelings - maybe it’s just because it’s August and beautiful, but the monthly edition of the PDX Health community meetup was a ripper. Make sure to reach out if interested to join for the next one. STAT’s Coverage of Epic This Month: Brittany Trang of STAT did a fantastic job of breaking the FTC investigation that’s been lurking, as well as UGM related coverage, so I had to give her her wish of a meme. Fall Conferences: I’m pumped to be kicking off my fall conference season in NYC at Nabla Accelerate next month, which Chrissy provided the link to apply for. Here’s the rest of my schedule in case you want to meet up: Commonwell (Redwood Shores, CA): Oct 13-14 Open@Epic (Madison, WI): Oct 21-22 eHealthExchange (Austin, TX): Oct 27 Sequoia Project (Austin, TX): Oct 28-29 HLTH (Vegas): Nov 15-18 RSNA (Chicago): Nov 29-Dec 3 Posts I Liked: HTI-6 Should Unbundle API Certification: Josh Mandel makes the case that HTI-6 should split (g)(10) into separate authorization and data certifications, so a PACS or a genomics platform can certify only the role it actually performs. Imaging is the urgent example, but the structure solves a much bigger problem. On implementing ePrior Auth with Epic: Scott Rossignol’s field notes from a live ePrior Auth build on Epic, including the two app registrations nobody warns you about and the CPT mapping problem waiting at the end. CMS-0057 implementations are going to be such a beast. Healthcare point solutions are starting to look a lot like streaming services: Spencer Dorn runs the cable-to-streaming arc against health IT and lands squarely on the bundling half of the cycle. Bundling can be good! It can also be bad! Life is nuanced. Health API Guy is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber. Subscribe
Alienware launches 32-inch 4K Tandem WOLED monitor for $699 and 560Hz QD-OLED model
Alienware has announced two RGB-stripe OLED gaming monitors, including a 32-inch 4K Tandem WOLED model priced at $699. The second model uses a 560Hz QD-OLED panel.
South Korea science ministry seeks record $21.6 billion budget
S. Korea's science ministry proposed a record 29.6 trillion won budget for 2027, with major increases for AI models and computing infrastructure.
Anthropic doubles a science benchmark score with Fable 5.1 while OpenAI says its Astra model crosses critical cyber threshold
Anthropic’s Fable 5.1 doubles a science benchmark score, while OpenAI says Astra reaches critical cyber capability and finds zero-day flaws.
Apple Watch Series 12: Three new features coming next week
Apple Watch Series 12 is coming next week at the September event, here are three new features rumored for the next-generation model.
Foldable iPhone Ultra Likely Has MagSafe Charging
Apple's foldable "iPhone Ultra" will likely support MagSafe charging, Bloomberg's Mark Gurman said in an interview with Tom's Guide. iPhone Ultra dummy models that surfaced in April didn't include a MagSafe ring, while dummies of the iPhone 18 Pro did. Leaked cases for the device did have a MagSafe ring, making it unclear whether it would support MagSafe charging. "All indications that I've heard is that MagSafe is in for this foldable phone.
The benefits of AI model ownership versus leasing
Reltio's Manish Sood discusses why companies should own AI loops instead of models, emphasizing proprietary data and continuous learning.
Meet Taylor Swift and Travis Kelce's new dog. She's already a model.
Taylor Swift and Travis Kelce’s dog stepped into the spotlight with a starring role in Kelce’s new Tommy Hilfiger campaign.
Perplexity's Hybrid Compute Splits Sensitive Tasks Between Cloud And Local AI
Perplexity's Hybrid Compute lets you split tasks between cloud and local models.
NVIDIA DLSS 5 Technical Preview Review
NVIDIA DLSS 5 is the first technology that adds to a scene rather than reconstructing it—a generative model running locally that magically improves things. In our DLSS 5 Technical Preview we look at how the model works, what it costs in performance, and how much control developers really have.
Study Identifies Potential Therapeutic Target Protein that Drives Metabolic Liver Disease Independent of Diet
The protein appears to cause liver damage by triggering an out-of-place antiviral immune response. Blocking this pathway can protect against liver injury in human cells and mouse models, even on a high-fat diet—paving the way for therapies to prevent liver damage.
LZ Sees Surprising Result in Search for Dark Matter
Key Takeaways Dark matter is the mysterious substance that accounts for the vast majority of matter in our universe. LUX-ZEPLIN (LZ) is a world-leading dark matter detector and specializes in searching for WIMPs, or weakly interacting massive particles. A new LZ study found one particle interaction that is difficult to explain with known background processes and could potentially have been caused by a WIMP. LZ continues to run to see if additional data strengthens or weakens the hint of a dark matter interaction. For the better part of a century, people have been trying to understand dark matter. This invisible substance makes up roughly 85% of the mass in the universe but has never been directly detected. Determining exactly what it is remains one of the biggest questions about our world. Now, a new analysis from the LUX-ZEPLIN (LZ) experiment has recorded a single particle interaction that researchers have great difficulty explaining with known background signals from normal matter. The result does not yet meet the statistical threshold required to claim a discovery, but is the most compelling hint of dark matter reported by the experiment to date. LZ is an international collaboration of 250 scientists and engineers from 39 institutions. The detector is managed by the U.S. Department of Energy’s Lawrence Berkeley National Laboratory (Berkeley Lab) and operates nearly one mile below ground at the Sanford Underground Research Facility (SURF) in South Dakota. The experiment uses 10 tonnes of ultrapure liquid xenon to search for dark matter and is optimized to look for WIMPs, or weakly interacting massive particles. The results were presented in a scientific talk at the 2026 TeV Particle Astrophysics conference in Japan. The paper will be released on the online repository arXiv and submitted to the journal Physical Review Letters. “We’re very intrigued to see this event in the data, in the region where we expect dark matter to show up and the competing backgrounds are very low,” said Rick Gaitskell, a professor at Brown University and the spokesperson for LZ. “With only one event, we don’t want to get ahead of ourselves. We are not claiming to have seen dark matter. But we have seen something interesting that we want to share with the scientific community for their input.” The LZ collaboration studies experimental data in batches. In the new result, researchers analyzed 220 live days of data collected between March 2023 and April 2024. The collaboration had previously searched this dataset for faint signals from the simplest kinds of WIMP interactions. The new analysis searched for a broader range of possible WIMP interactions that could deposit more energy in the detector. LZ is particularly sensitive to such signals while also minimizing false positives. “This was a detailed study in a region we hadn’t explored within this dataset, and we spent months of additional effort to understand all the possible causes of background events,” said Sam Eriksen, a senior research associate at the University of Bristol in the U.K. and lead author of the study. “We understand our detector and the backgrounds so well that even a single outstanding event, like the one we found, is important. We expect dark matter events to be extremely rare, so only a handful could mark the first detection of WIMP dark matter.” If the anomalous event was caused by dark matter, the WIMP that generated it would likely have a mass of at least 200 GeV/c2 (gigaelectronvolts), or more than 200 times the mass of a proton. It would also suggest a specific type of interaction between WIMPs and ordinary matter beyond the simplest model. The LZ results have not reached “5-sigma” significance, the statistical threshold considered a discovery in physics. The new analysis is 2.6 sigma, meaning there is approximately a 0.5% chance that the event could be explained by known backgrounds. With additional data, researchers can test whether the finding continues to grow in significance or fades away. LZ has already accumulated the world’s largest dark matter dataset and will continue to accrue WIMP search data at SURF, substantially improving their search statistics. LZ searches for dark matter by looking for signature flashes of light from energy deposited in the detector. The collaboration leverages multiple methods to prevent or account for particle interactions caused by normal matter. This includes the mile of rock that shields the detector from cosmic rays from space, a water tank and outer detectors that protect the central detector from background neutrons, and a suite of computational tools that disentangle particle interactions and reject dark matter mimics. “Outlier events in the data are not unexpected, but they usually stand out as a background of some kind when you look at them deeper,” said Aaron Manalaysay, a physicist at Berkeley Lab and the chair of LZ’s Institutional Board. “This is the first example in any experiment I’ve worked on of an outlier that appears valid in every way. Of course, we’re still twisting our brains trying to think if there’s a rare background mechanism we could’ve missed, but it’s thrilling to wonder if this could be the first hint of a dark-matter observation.” LZ is supported by the U.S. Department of Energy, Office of Science, Office of High Energy and Nuclear Physics, and the National Energy Research Scientific Computing Center, a DOE Office of Science user facility. LZ is also supported by the Science & Technology Facilities Council of the United Kingdom; the Portuguese Foundation for Science and Technology; the Swiss National Science Foundation; the Australian Research Council Centre of Excellence for Dark Matter Particle Physics; and the Institute for Basic Science, Korea. Thirty-nine institutions of higher education and advanced research provided support to LZ. The LZ collaboration acknowledges the assistance of the Sanford Underground Research Facility. ### Lawrence Berkeley National Laboratory (Berkeley Lab) is committed to groundbreaking research focused on discovery science and solutions for abundant and reliable energy supplies. The lab’s expertise spans materials, chemistry, physics, biology, earth and environmental science, mathematics, and computing. Researchers from around the world rely on the lab’s world-class scientific facilities for their own pioneering research. Founded in 1931 on the belief that the biggest problems are best addressed by teams, Berkeley Lab and its scientists have been recognized with 17 Nobel Prizes. Berkeley Lab is a multiprogram national laboratory managed by the University of California for the U.S. Department of Energy’s Office of Science. DOE’s Office of Science is the single largest supporter of basic research in the physical sciences in the United States, and is working to address some of the most pressing challenges of our time. For more information, please visit energy.gov/science.
AI Agents Are Learning to Skirt the Rules. Can Businesses Keep Them Under Control?
“Companies have no reliable way to know how a model will behave under pressure.” That concern inspired SnowCrash Labs, an AI safety startup with Darden alumni and student connections.
Hyundai Turns Performance Into a Bigger Business
(TestMiles) – Hyundai plans to grow N beyond seven models while Genesis Magma reaches for luxury buyers, turning speed into a broader brand strategy. Hyundai has spent years proving it can ma…
Samsung's New Galaxy Book 6 Model Is A Budget-Friendly $799 Laptop
Samsung has released a more accessible and affordable Galaxy Book 6 laptop.
Anthropic tightens training security after Claude agents went rogue
Anthropic tightened its AI security and paused some programs after Claude models accessed three organizations' systems without permission in April.
Business News September 2026
• Fall at the Farmers Market • The Fraud Clock is Ticking • Remodel to achieve your Forever Home • Roger, the Computer Guy! • Cornell Farm’s Century Farm Celebration • Grand Opening Celebration at OR Credit Union • The Confidence Within: an educational pelvic wellness experience • Back-to-School Dinner Dilemma, Solved
Why Canada-US Risk Analysis Should Start With The Business Model
Convetional risk analysis often fails to identfiy the extent which government policy will affect the ecnomic value of the company.
Stability AI Raises $76M Series B, Backed by Major Entertainment Companies
Stability AI, the startup behind image generation model Stable Diffusion, has raised $76 million in Series B funding, bringing its total funding to $232
Run NVIDIA BioNeMo NIM Microservices for Protein Structure Prediction in Claude Science
Agentic AI is changing how research is done. AI scientists can read papers, propose hypotheses, call models, and determine which experiments to prioritize next. First proving their value in software…
iPhone 18 Pro is coming: Here’s what’s new with each model
iPhone 18 Pro will be unveiled next week, with new Pro and Pro Max models debuting. Here’s what rumors say is new with each model.
Model Helps Scientists Build Better Protein-Degrading Therapies
A team of Weill Cornell Medicine investigators has developed a computer model that will help scientists more efficiently develop protein-destroying therapies, a newer type of treatment that is particularly promising for cancer.
Advanced AI threatens the world’s financial system, watchdog says
The head of the world’s financial stability watchdog warned on Monday of the serious risk posed by advanced AI models to the global financial system.
Seoul National University of Science and Technology develops AI framework to test and improve vision-system robustness
A new artificial intelligence (AI)-based system shows how natural damage to traffic signs can mislead deep neural network (DNN)-based vision systems. The framework, AdvWT, uses a generative model to learn realistic damage patterns and then optimize them to trigger misclassification. Tests across multiple AI architectures showed strong attack success, transferability and physical-world robustness. The approach can identify vulnerabilities and improve AI generalization to real-world damage.
Global Social Entertainment Market Consolidates Around Three Leaders, with Newborn Town Emerging as a Key Challenger
EQS Newswire / 31/08/2026 / 15:26 UTC+8In 2026, the global social entertainment industry is undergoing a significant reshaping. The growth model is shifting from broad-based user acquisition to deeper regional expansion, and from one-size-fits-all products to differentiated strategies tailored to s…
AI streamlines search for new ceramics that generate electricity
To streamline the search for new piezoelectric materials, or those that emit an electric charge when manipulated, a team led by materials science researchers at Penn State introduced a new framework that combines an existing artificial intelligence model with human expertise and experimentation.
Lummi Nation offers model for county's path to behavioral health, justice
Appropriate behavioral health care should be at core of mission to deliver justice
Antarctica developed a continental-scale ice sheet 34 million years ago while Earth was about 5°C warmer than today — helped by slow-moving mantle waves that had spent more than 100 million years lifting its interior high enough for permanent ice to survive.
A 2026 modelling study links East Antarctica’s high plateau to mantle processes begun during Jurassic breakup, while retaining falling carbon dioxide as the main trigger for glaciation.
Health taxes: Overcoming technical and political challenges, including illicit trade concerns, and lessons from country experience
A virtual side event ahead of the seventy-seventh session of the WHO Regional Committee for the Western Pacific will explore how health taxes on tobacco, alcohol and sugar-sweetened beverages can reduce unhealthy consumption, generate revenue and support health priorities. Aligned with WHO’s 3 by 35 Initiative, the session will share country experiences in tax modelling, revenue allocation and addressing illicit trade, while highlighting practical strategies for advancing health tax policies. The event will be livestreamed on X, Facebook, YouTube and the RCM77 web page.
Apple Caught Off Guard by AI Demand for Mac Mini and Mac Studio
Apple's unusually timed announcement of new Mac mini and Mac Studio models this week was driven by unexpectedly strong enterprise appetite for AI hardware, according to The Information. Apple normally releases new Mac models in the autumn, closer to October or November, making this week's announcement unusually early, falling just before the anticipated arrival of new iPhone models. The Information says that the AI-driven boom in Mac Studio and Mac mini sales is behind the early launch.
New Apple Watch Series 12 and Apple Watch Ultra 4 Features Revealed
Bloomberg's Mark Gurman today further outlined what to expect from the Apple Watch Series 12 and Apple Watch Ultra 4 set to be unveiled on September 9. While it was initially rumored that Touch ID may be coming to the Apple Watch this year, Gurman reiterated that he does not expect the Series 12 or Ultra 4 to have any major design changes compared to the current Series 11 and Ultra 3 models. Gurman does expect a ceramic case option to return for the Apple Watch Series 12.
Scientists Are Using AI And Viruses To Help Defeat Superbugs
Deep learning models can help scientists build customized phages.
Scientists Are Using AI And Viruses To Help Defeat Superbugs
Deep learning models can help scientists build customized phages.
Colorado Sues EarnIn For “Immoral, Unethical, Oppressive and Unscrupulous” Business Practices
Hey all, Jason here. When this hits your inbox, I should be en route to the airport — not to go anywhere myself, just to pick up my better half. I must admit, it’s been strange to be home alone (well, with two dogs) for two weeks! Really lets me “optimize” to spending more time on reading/writing/working, whether that’s healthy or not! Looking forward to getting back into a (somewhat) more balance routine. Subscribe or Support by Upgrading Compare Notes With the People Who Built Modern Fintech Partner content: Remitly CEO Sebastian Gunningham. Increase founder Darragh Buckley. Linda Du, Co-founder & President at Valon. They’ll unpack the decisions behind building and scaling major financial products: what they got right, what broke, what they would do differently, and how AI and new infrastructure are changing the operator playbook. Most events tell you where fintech is going. NerdCon puts the people who built the present in the room and asks them what comes next. Expect an operator-led conversation grounded in real products, real trade-offs and lessons earned the hard way. Fintech Business Weekly readers save 20% on Regular tickets with code FBW20 before standard pricing ends September 11. Meet the Builders Things To Know & Other Good Reads Advent and Stripe Abandon $50 Billion Pursuit of PayPal (Bloomberg) Trump Family’s New Crypto Bank Is Backed by Abu Dhabi Sheikh (Wall Street Journal) The multiplying risks of financing data centres (FT) Let the Bond Market Speak (Wall Street Journal) Did we waste a crisis? Modest proposals to reform deposit insurance. (Fintech Takes Banking) Untangling Guggenheim: How Private Credit Built Its Own Universe (Net Interest) The Hater’s Guide To Circular Financing — Part One (Ed Zitron) Fiat Ventures Rebrands to FGV Capital, Announces $35M Oversubscribed Fund II (BusinessWire) Socure Announces Strategic Growth Investment at $5.2B Valuation and Acquires Agentic Operations Platform Fravity (Socure) Listen: When Consumer Protection Disclosures Work Too Well (Consumer Finance Monitor Podcast) Colorado Sues EarnIn For “Immoral, Unethical, Oppressive and Unscrupulous” Business Practices When Andreessen Horowitz led EarnIn’s $39 million Series B in 2017, Alex Rampell, a partner at the storied venture capital firm, described EarnIn’s small-dollar advance product as “free,” relying on “entirely on voluntary contributions instead of fees.” Rampell favorably compared EarnIn’s approach to typical payday loans, writing, “Payday loans historically have had the potential to be a slippery slope for consumers into financial distress: opaque systems and steep fees that are hard to repay and set individuals back further than where they started. For many Americans living paycheck to paycheck, that kind of slide into debt can be extremely difficult to recover from.” But EarnIn’s product — which, the company says, isn’t a “loan,” legally speaking — ends up being anything but free for most users. While it is technically possible to take a no-fee advance from the company, most users incur “Lightning Speed” (instant funding) fees and/or ostensibly optional tips that combined can equate to annualized percent rates that can reach over 1,000%. With an average term of 9.77 days, even fees and tips that are small on an absolute dollar basis are equivalent to high APRs on an annualized basis. A lawsuit filed last week by Colorado Attorney General Phil Weiser against EarnIn describes the company and its product and business practices quite differently than Andreessen’s Rampell; “EarnIn’s business practices are unfair because they are immoral, unethical, oppressive and unscrupulous,” the complaint, filed in District Court in Denver, Colorado, says. EarnIn’s marquee product, Cash Out, isn’t a loan, the company says, as Cash Out is not “[f]orward [l]ooking” because “[t]he money is transferred based on earnings to date,” does not carry a mandatory fee to access funds, and is “non-recourse,” meaning EarnIn cannot pursue recovery from users who don’t repay advanced funds. EarnIn is sometimes referred to as an “Earned Wage Advance” service. Though, unlike some others in the broader category, EarnIn does not integrate with employers’ payroll or time and attendance systems. Rather, EarnIn attempts to verify a user’s income and employment, by examining direct deposit transaction data in their linked external bank account, by using GPS location and Bluetooth data to estimate a user’s time spent at work, if they work in a fixed location, and/or by having a user provided a work email address. While repayment is theoretically optional, as EarnIn would have no recourse, such as reporting non-payment to the credit bureaus or a collections agency, this is anything but clear to users of the EarnIn app, Colorado’s complaint argues. During the process of taking a Cash Out, users must agree to a preauthorized debit of the amount advanced, plus any expedited funding fee and/or tips. Though users technically can revoke this ACH authorization, the fact that they have this right isn’t presented to users during the Cash Out process, but rather is included in linked terms and conditions few users are likely to actually read. And users who do wish to revoke their ACH authorization must do so three or more days before the scheduled transaction date by emailing EarnIn’s customer support. Similarly, EarnIn’s marketing claims that it carries “no interest” and “no mandatory fees” is technically true, but is not the reality for many of the app’s users, the Colorado complaint argues. Further, EarnIn has extensively marketed to users the ability to “access your pay today,” to get “instant” funds, and to get funds “the same day you work” — when, in reality, this speed of access was only possible by paying the additional, undisclosed Lightning Speed fee. The Colorado complaint highlights this contradiction specifically, noting: EarnIn tells consumers that they can ‘Make any day payday with EarnIn’ and that there no ‘hidden’ or ‘mandatory fees.’ For a consumer to access their pay on the same day requested, they would have to pay the Lightning Speed fee. In the fine print of the advertisement EarnIn discloses that ‘[f]ees apply to use Lightning Speed,’ a direct contradiction to the larger text of the advertisement, and a fact that would not be immediately understood by a consumer as ‘Lightning Speed’ is not defined nor explained. Like other small-dollar lending services, rather than explicitly charge interest, EarnIn collects a markup on the optional Lighting Speed fee, which, given the nature of the transactions, many users opt for, as well as offering users the option to leave “tips.” EarnIn leverages either Real-Time Payments (RTP) rails or push-to-card via debit rails if a user opts for the optional Lightning Speed. But while these capabilities cost EarnIn, on average, $0.075 or $0.20 per transaction, respectively, EarnIn charges users significantly more. Initially, EarnIn charged as little as $1.99 for Cash Outs up to $24. But EarnIn has hiked these fees multiple times, and now charges $4.99 for Cash Outs up to $75 and $6.99 for those over $75. According to the Colorado complaint, EarnIn leverages “dark patterns” to manipulate users and to make it more difficult to avoid tipping. Examples of these “road blocks” highlighted in the complaint include: As of 2023, in the EarnIn app, a consumer attempting to get a loan under the default settings had to make at least eighteen separate taps to complete the transaction and reduce the tip to $0. There is an alternative method to leave a $0 tip, via the “Custom tip” feature, but EarnIn made this option difficult to access by visually deemphasizing it and positioning it near devices’ “home” button. Using language and imagery such as “Tip to pay it forward,” “Your generosity supports the service + helps,” “Your tips make a difference,” and “[t]he APR for this cash out is 0%. Your tips help support us.” While Lightning Speed and tipping are, in theory, both optional, in practice, users nearly always ended up paying something. According to the Colorado suit, users were charged either a tip or an expedite fee for more than 92% of transactions, with an average APR of nearly 388%. And, although the advances are “non-recourse,” EarnIn successfully collects on nearly all transactions — 99.18%, according to the complaint, which, Colorado argues, “demonstrat[es] that EarnIn’s advances function as high cost loans, not voluntary payments.” The share of EarnIn users incurring fees — despite the product being marketed as 0% APR — is hard to reconcile with Andreessen investor Rampell’s favorable description of EarnIn vs. “opaque systems and steep fees” associated with payday loans, which, Rampell says, can be “extremely difficult” for borrowers to extricate themselves from. According to Colorado’s suit, EarnIn’s Cash Out product “trapped many consumers in extreme cycles of high-cost reborrowing.” The suit gives an example of one consumer in the state, who took out a whopping 1,151 loans, paying a total of $4,038.50 in Lightning Speed expedited funding fees on loans that averaged the equivalent of 1,421% APR. Another Colorado user, the suit says, took out 1,033 loans, paying $8,561.22 in tips and fees at an average APR of 1,539% . These types of usage patterns demonstrate “the severe and repeated financial harm caused by EarnIn’s illegal lending model,” the Colorado complaint argues. In aggregate, from January 2023 through July 2025, EarnIn extended 3.1 million loans to 56,778 users in Colorado — an average of nearly 55 transactions per person during the time period. EarnIn lent more than $300 million and collected more than $16 million from Colorado users in Lightning Speed fees and in tips. Ultimately, the Colorado complaint argues that “EarnIn’s Cash Outs are loans under Colorado law, and EarnIn’s contention that its product is not a loan based on its disclaimer of any legal repayment obligation lacks any real-world significance given how the Cash Out transactions actually operate in practice.” Colorado further argues that “EarnIn was not working with companies here in providing consumers with funds but acted as a third-party lender and charged illegally high rates, used deceptive design strategies to extract some charges, and trapped consumers in repeat borrowing. Colorado will continue to stand up for consumers and hold companies accountable when they violate our credit laws or attempt to evade them through misleading practices.” Colorado’s suit alleges EarnIn assesses finance charge in excess of that permitted by Colorado law, that EarnIn failed to make legally required disclosures to Colorado borrowers, that EarnIn made supervised loans to consumers without obtaining the requisite license, that EarnIn violated the state’s Deferred Deposit Loan Act, that EarnIn engage in unfair and deceptive trade practices, and that EarnIn made false or misleading statements concerning price. The suit seeks a court order enjoining EarnIn from violating relevant Colorado laws, for EarnIn to refund amounts charged in excess of state law, fees, costs, and penalties, and other relief the court deems to be just. Fintech Business Weekly is made possible by the generous support of paying subscribers — bringing you independent analysis of banking, fintech, and crypto without fear or favor. You can support my work by becoming a paying subscriber if you aren’t already. Paying subscribers enjoy: access to the entire archive of nearly six years of newsletters extended versions of the weekly newsletter, with additional content and analysis and (for founding member tier) quarterly personal 1:1 fintech Q&A / strategy calls with me, tapping into my unparalleled knowledge of the intersection of banking and fintech and experience working in the sector, including helping to launch Goldman Sachs’ retail bank Marcus Support Fintech Business Weekly You can also support Fintech Business Weekly by sponsoring a newsletter or podcast, putting you in front of 93,000+ of the most influential decision makers in banking, fintech, and crypto. Learn more about sponsor opportunities or request a media kit by dropping me an email. Q2 Quarterly Banking Profile: Five Quick Takeaways The FDIC released its quarterly banking profile, which is always worth a read, for the second quarter. The profile provides an aggregate snapshot about the overall health of the 4,238 institutions whose deposits are insured by the FDIC, who, collectively, hold more than $26 trillion in assets and saw a total net income of more than $90 billion in the second quarter. Net income was up 12% vs. Q1 2026, with lower provisions and realized gains on securities helping to drive the improvement, somewhat offset by increases in non-interest expense and applicable income taxes. Overall net interest margin was basically flat quarter over quarter (up 1 basis point vs. Q1 2026), though community banks saw NIM grow by 10 basis points, to 3.81%. Unsurprisingly, given that interest rates haven’t changed, unrealized losses on securities changed little, edging up slightly to $326.7 billion vs. $325.1 billion in Q1 2026. Aggregate credit quality has remained benign, with asset quality actually improving slightly in Q2. Overall, institutions reported a 9 basis point decline in their past-due and non-accrual rate, to 1.44%. Net charge off rate also declined by 2 basis points, to 0.57%. And finally, the FDIC’s deposit insurance fund stands at $161.1 billion, making the DIF reserve ratio 1.48% — above the statutory minimum of 1.35%, but still below the designated target reserve ratio of 2%. [Paid Subscriber Exclusive] Nissan Withdraws ILC App, State Associations Plan BankChain Alliance, FDIC Defines “Unsafe or Unsound Practice” For First Time Automaker Nissan withdrew its application to charter a Utah industrial loan company and its corresponding application to the FDIC for deposit insurance last week, on August 19th. Nissan’s application had been outstanding for more than a year, as it originally applied in June 2025. Recent applications to the OCC to form national banks have been decisioned, one way or the other, substantially more quickly than has historically been the case.
A fifth-grader in Kansas City clicked ball-and-stick atoms together in science class and asked her teacher if she'd made a real molecule — he photographed it, sent it to a chemistry professor, and at ten years old she became co-author of a published paper on a structure nobody had ever described
A ten-year-old's casual question about her chemistry model kit led to a published paper describing a molecular structure that had eluded professional scientists for decades.
Top Stories: Apple's 'Surprise and Shine' Event, Plus New Mac mini and Mac Studio
It turned out to be a big week for Apple news this week, as we not only got an official announcement about the upcoming iPhone event but also several hardware announcements including updated Mac mini and Mac Studio models. Anticipation continues to build for Apple's first foldable iPhone coming at next month's event, while subscribers to the Apple TV streaming service are getting hit with another price increase, so read on below for all of the details.
Amazon Offers Early Savings on 2026 Mac Mini, Plus Record Low on AirTag 2
Apple introduced the M6 and M5 Pro Mac mini this week, and Amazon currently has the first pre-order discounts on these models ahead of their September launch. Below, we're also tracking a new record low price on the AirTag 2 during Amazon's Labor Day event. Note: MacRumors is an affiliate partner with some of these vendors. When you click a link and make a purchase, we may receive a small payment, which helps us keep the site running.
How to Run a Chatbot on Your Own Computer
Installing a large language model on your personal computer gives you a handy digital assistant that won’t compromise your data privacy.
How the Cold War US Military Modeled Itself on Private Business
The following is an excerpt from Pentagon Capitalism: How the Cold War US Military Modeled Itself on Private Business by A. J. Murphy, now out at Harvard University Press.
iPhone 18 Pro Pre-Orders May Start at an Unusual Time
Pre-orders of Apple's iPhone 18 Pro may not open until Saturday, September 12, a day later than the company's usual schedule, according to a new report. Apple is set to hold its annual iPhone event on Wednesday, September 9, where the iPhone 18 Pro models, its first-ever foldable iPhone, and the Apple Watch Series 12 and Ultra 4 are all expected to be announced.
What Not to Expect at Apple's 'Surprise and Shine' September 9 Event
Apple's annual fall event takes place on Wednesday, September 9, when it is expected to unveil new iPhone 18 Pro models, the first foldable iPhone, an Apple Watch Series 12, a fourth-generation Apple Watch Ultra, and potentially AirPods 5. Apple may also make some additional new products available at its "Surprise and shine" event, but there are a handful of rumored upcoming devices that we don't expect to make an appearance this month.
Apple's M6 Mac Mini Gets First Pre-Order Discounts at Amazon
Amazon today has introduced the first cash discounts on Apple's brand new Mac mini, with discounts including two of the M6 models as well as the M5 Pro model. All of these deals are pre-order discounts on the 2026 Mac mini, which officially launches on September 22. Note: MacRumors is an affiliate partner with Amazon. When you click a link and make a purchase, we may receive a small payment, which helps us keep the site running.
To match Denmark’s public health, we need more than its vaccine model
There are plenty of lessons the U.S. should take from countries such as Denmark, but vaccine policy is not one of them.
How batteries are reshaping the solar power business model
Also in today’s newsletter, are global emissions approaching a peak?
Aldine Students Turn Food Truck Models Into Lessons in Business, Teamwork and Entrepreneurship
IMAGE For students in Marilyn Johnson’s business classes at Eisenhower Ninth Grade College Prep, a food truck begins with more than a menu. It starts with a question: What do their customers want? Over the past four years, Johnson has turned a classroom assignment into a hands-on business simulation that asks students to take an idea from customer research to a 3D food truck model — and build a business around it. The project recently received a boost when Greater Texas Credit Union presented Johnson with a $250 award recognizing her commitment to students and meaningful learning experiences. “I want students [...]
The Literacy Crisis Reaches Dog-Bear Science
“New analysis of more than 300 bear encounters suggests our canine companions often increase odds of conflict and rarely serve as early warning systems many recreationists assume,” reads a subhead in Mountain Journal. The thing is, that study reaches no such conclusion. But you have to read beyond the abstract to understand that. Mountain Journal is not alone. Let’s look at a brief list of takeaways published elsewhere just to establish that none of them seem to know how to read science. Wes Siler’s Newsletter is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber. Subscribe Here’s CBC: “A new peer-reviewed study, The Canine Conundrum: Is a Dog a Help or Hindrance in Bear Country?, dug into conflicts similar to what the video depicts, by analyzing 326 reported human-bear conflicts from 1901 to 2023 involving dogs.” “The study found 54 per cent of conflicts were started by dogs. In the encounters researchers included in the study, five people and 56 dogs were killed, and 170 people were injured.” “Lana Ciarniello, an independent researcher who co-authored the study, said researchers found when a person goes hiking with at least one other person, the odds of a negative bear encounter decrease by 50 per cent.” “However, when a dog is involved, the chances of a negative bear encounter increase because ‘a dog really adds a chaotic factor into the mix of recreation,’ said Ciarniello.” Again, the study finds no data supporting that conclusion. “Hiking with your pet dog may make you feel safer in bear country—but a hiker’s recent encounter and a new study prove that the numbers don’t add up,” writes Backpacker, which is now part of Outside. The numbers don’t add up to support this take. There’s plenty more examples, but all are equally dumb, so I’ll stop there. How did so many people get this so wrong? Three reasons: 1) the study is published behind a paywall, with only the abstract freely available. 2) the methodology sets out to create work relevant to a hyper-specific audience, not the public. And 3) the dataset studied is almost absurdly limited and doesn’t bear (get it?!) comparison to wider information. Let’s work through all three. The Problem is Money I understand the irony, since I’m writing about this on a subscription-model newsletter, but information wants to be free. The study is published by The Wildlife Society and costs $16 to access. Reprinting it is specifically prohibited by the terms of use, so I can only share snippets and my conclusions with you here. Having worked as a journalist for 24 years, I can report with some authority that journalists are both lazy and cheap. I don’t think it’s a stretch to say that much of the material you’re seeing written about this study is drawn from the freely-available abstract (above) only, and not a deep dive into its methodology or data. And that’s a problem, because abstracts are where researchers are able to break from their numbers and editorialize their opinions about conclusions. So what’s getting reported isn’t based on numbers, it’s based on an opinion designed for a specific audience, not the general public. Hell yeah ladies, kick that bear’s ass. The Methodology is Designed to Support a Predetermined Conclusion One of the biggest issues with the science that everyone likes to cite about bear spray’s efficacy is that no one understands that study was designed for a specific audience, looking for a specific answer. And since that audience is not the public, and our safety was not the question being asked, its conclusion is rendered irrelevant to our needs. Yes, I’m talking about Efficacy of Bear Deterrent Spray in Alaska, which was created by Tom Smith and Stephen Herrero, two of the same scientists who put together this dog-bear conflict analysis. In the spray study, the purpose was to present relevant agencies with the information they needed to issue their field workers with bear spray. As such, incidents where people were unable to access the spray or similarly failed to get it deployed in time were excluded from the results, and only incidents they deemed successful were included. Could a ranger roll down their vehicle’s window and scare a bear away from a dumpster by spraying it? Turns out the answer was yes, and that was good enough for the purposes of that study. The sample size of actual bear attacks—the thing you and I would want bear spray to provide a solution to, was too small to draw conclusions from, and didn’t indicate success in that role. Is a Dog a Help or Hindrance in Bear Country? has similar limitations. “Insights provided here are intended to help wildlife managers create effective guidelines regarding the presence of domestic dogs in bear habitat,” is explained clearly by the study’s text. To conduct the study, Smith and Herrero went looking for publicly available reports of dog-bear conflicts: “All incidents included in our analysis involved at least one bear, one dog, and one person who was at immediate risk of injury (e.g., not in a protective structure such as a cabin or vehicle), injured, or killed,” explains the report’s methodology. Why is that a problem? “We recognize that data used in this study are biased in that people generally only report adverse interactions with bears, not benign ones. Consequently, we assume that for the most part, people and their dogs pass through bear country without incident.” As was the problem with the bear spray study, people are drawing conclusions here from a study that was never intended to be relevant to the general public. The Smith-Herrero database famously excludes Alaska’s Defense of Life and Property records, which are elsewhere considered the largest and most authoritative source on human-bear conflict. The Dataset is Absurdly Limited A 2009 study conducted by the Washington Department of Fish & Wildlife estimates that each year there are at least 43,000 documented complaints of human-bear conflict across North America. Is a Dog a Help or Hindrance in Bear Country? Examined those reports for years ranging from 1901 to 2023, and only managed to find 326 such incidents that met the criteria explained above. Let’s do some quick math, poorly. 43,000 incidents x 122 years = 5,246,000 total incidents of documented human-bear conflict. Smith and Herrero were only able to find 326 incidents that met their criteria for dog+human+bear+risk or occurrence of injury. That means there’s a 0.006 percent chance that the study’s conclusion—that dogs make bear encounters more dangerous—even gets a chance of occurring. And remember, that’s only in documented conflicts. There’s no way to know how many people across North America run into a bear in any given year without any conflict, but it’s probably safe to assume that the total number is well in excess of 43,000. And, dig deeper into the study and you can find further restrictions that skew its conclusions even more. “While the people involved in these incidents did not report the level of training their dogs had received, it is our assumption that most were untrained companion dogs as opposed to well-trained bear dogs, such as Karelians and livestock guardian breeds that have proven to be significant aids in bear country,” the study’s methodology explains. And that’s a problem, because an ongoing study here in Montana is finding that the simple presence of a Kangal like our youngest dog Teddy can reduce visits to private property from grizzly bears by 87.8 percent. In addition to being a badass, Teddy is also a companion animal. And in my admittedly limited sample size of the three dogs currently present in our living room, she’s no more effective than her mutt brothers. Plenty of people own a dog capable of deterring bears, even if they didn’t set out to own a bear dog. The real purpose of this study appears to be a recommendation that leashes are mandated and recreationists pursue training in their deterrent of choice. But that doesn’t make for sensational headlines. Are There Any Real Takeaways Here? Let’s throw some numbers around, just to try and get our hands around what the larger picture might look like for bears and dogs here. The American Veterinary Medical Association estimates that 42.6 percent of American households own a dog. The Canadian Animal Health Institute says roughly 35 percent of households in Canuckistan own a dog. In an attempt to create the worst possible result for dogs, let’s run with that 35 percent number. And given that outdoorsy and rural people are more likely to both own a dog, and have bear encounters, let’s just make a giant assumption and apply that number to those 43,000 conflicts documented annually. If 35 percent of 43,000 human-bear encounters involve a dog, then we can assume that at least 15,050 dog-bear conflicts are occurring annually in North America. 15,050 x 122 = 1,836,100. Smith and Herrero were only able to find 326 incidents that met their human, dog, chance or occurrence of injury criteria across 122 years. And only 170 documented human injuries. So, with all the obvious caveats about these numbers in no way describing reality beyond narrow study parameters, what they actually found is that your risk of injury if your dog tangles with a bear is only 0.009 percent. I’ll take those odds next time I’m hiking my dogs in bear country. Top photo: “Superintendent Horace Albright, eating pancakes with bears, at Lake Lodge, Yellowstone National Park.” NPS. A journalist with more than two decades of experience working around the world, Wes Siler is here to cut through the outrage and disinformation to bring you the factual, insightful, actionable reporting you need to understand what’s going on. Upgrading to a paid subscription supports this reporting, and buys personal access to Wes, who will help you save money on gear, plan outdoor adventures, and prepare for real life, and who promises he’s less salty in real life than he sometimes comes across as on the Internet. Wes Siler’s Newsletter is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber. Subscribe
Anthropic makes first move into physical AI with new way for scientists, manufacturers to bring equipment to life
The Model Hardware Standard (MHS) works with any AI model, not just Claude, but for now only a handful of companies can access the research preview.
2025 vs. 2026 Mac Studio Buyer's Guide: All Upgrades Compared
Apple this week announced a new Mac Studio with the M5 Max and M5 Ultra chips, discontinuing the M4 Max and M3 Ultra models that had been on sale since March 2025. The machine's design is unchanged for a fifth consecutive year, with all of the upgrades being internal. The table below details everything that changed regardless of whether you choose a Max or an Ultra chip. The two tables that follow break down the differences within each tier.
College football odds, picks, predictions for Week 0, 2026: Computer backs Virginia, New Mexico State
SportsLine's model has revealed picks for every Week 0 college football game after 10,000 simulations
Apple's New Macs to Ship With macOS Golden Gate
The new Mac mini and Mac Studio models announced by Apple this week are "slated to ship" with macOS Golden Gate, according to Daring Fireball's John Gruber. Apple's product pages for the new Mac mini and Mac Studio models mention macOS Golden Gate features such as Siri AI, providing further evidence that the computers will ship with the latest operating system, instead of macOS Tahoe.
To match Denmark’s public health, we need more than its vaccine model
There are plenty of lessons the U.S. should take from countries such as Denmark, but vaccine policy is not one of them.
Samsung Introduces Three New Odyssey Series Gaming Monitors at Gamescom 2026
Samsung Electronics today announced its 2027 Odyssey gaming monitor lineup at Gamescom 2026, the world's largest gaming event, being held in Cologne, Germany from Aug. 26-30. The new lineup introduces multiple Odyssey models that feature world-first innovations, empowering players with greater...
Some Scientists Have ‘Magic Hands’ in the Lab. This A.I. Is Learning Why.
Even the most adept researchers may not know exactly what they do to get successful results. An A.I. model is trying to figure it out by watching every move.
Opinion | How Shenzhen can win the ‘AI for science’ race
Rather than copy the research-focused models of Beijing and Shanghai, the tech hub must leverage its industrial depths to become a premier lab for experimentation and deployment.
China's Internet celebrates Z.ai claiming ownership of Ox Alpha
Chinese fans rejoiced at Z.ai claiming ownership of Ox Aloha, lauding the startup's new model and its use of domestic AI chips.
Bitcoin treasury companies shed $80bn in value as business model unwinds
Crypto hoarders have been selling tokens and reverting to former activities
The Global Brain Care Coalition: A Model to Put Brain Science into Action
It takes an average of 17 years for a scientific discovery to travel from the lab to clinical practice. In brain health, that delay has a real cost as the globa
Bazzell on Unrivaled’s Five Year Plan
Alex Bazzell on Unrivaled's new $650 million valuation and player equity model.
Some assembly required, with machine learning
WashU researchers are working on ways to use machine learning to not just make predictions about potential chemical compounds but also to model out the synthesis of those materials.
Google launches Gemini 3.5 Transcribe, which powers Gboard Rambler & is coming to Chrome
Google today introduced Gemini 3.5 Transcribe as its “most precise speech-to-text model yet” that is already powering...
Galaxy S26 Ultra is the world’s most popular Android phone, and that’s a first
Android phones are never the top-selling smartphones globally, but the ones that get closest are almost always budget models. In...
AOC AGP327KG is the first 5K G-SYNC Pulsar gaming monitor
AOC has unveiled the AGP327KG, a 31.5-inch 5K gaming monitor running at 144Hz. It is the first model to combine 5120×2880 resolution with NVIDIA G-SYNC Pulsar.
Apple upgrading recent Mac mini orders to M6, M5 Pro models for free
Apple launched a new Mac mini with M6 and M5 Pro chips, and reports online show that recent M4 purchases are being upgraded for free.
New Mac Mini Bumps Display Refresh Rates to 165Hz
Apple's newly announced M6 and M5 Pro Mac mini models still support up to three external displays, but there's now support for higher refresh rates, according to Apple's published technical specifications. Here's what's changed versus the M4 generation: The M6 model runs two displays at up to 6K 60Hz or 4K 165Hz, up from 144Hz on M4, plus a third display at up to 5K 60Hz over Thunderbolt or 4K 60Hz over HDMI.
China's short-drama producers flood the market with cheap bets — and let audiences pick the winners
Short dramas' model posits a new alternative to traditional entertainment models, but distribution poses rising cost and audience retention challenges.
AI helps design new materials that work in the real world
MIT researchers added a new component to AI models that design new materials, helping ensure the material will be stable and practical for real-world use. The “CrysVCD” approach could reduce the money and time spent on screening out unusable designs.
Scientists Reconstruct Sound of Jurassic Insects
Using lasers, computer models and AI, researchers reconstructed the calls of 165-million-year-old relatives of modern crickets and katydids from fossilized wing fragments found in China.
Apple to Release 13 More New Products Later This Year
Looking past the new Mac mini and Mac Studio models unveiled today, Apple still has plenty of other new products on its roadmap for this year. It is nearly September, and that means the iPhone 18 Pro, iPhone 18 Pro Max, and the first-ever foldable iPhone are just around the corner. Other products expected later this year include the Apple Watch Series 12 with a ceramic case option, AirPods 5, an all-new Apple home hub, new Apple TV and HomePod mini models, and more.
QXO (QXO) is Rebuilding Business Models for Greater Value
Orbis Investment Management, an investment management company, released its Q2 2026 investor letter for “Orbis Global Equity Strategy”. The letter can be downloaded here. In the first half of 2026, the Global Equity Strategy returned 19.9%, surpassing the MSCI All Country World Index by 7.8%. The market faced narrow breadth in the quarter. The strategy focuses […]
First Look: The Updated Longines Master Collection now gets Light Blue Dials
Longines' recent love for ice blue gets spread on the Master Collection, with new Light Blue dials for th 39mm and 41mm automatic models.
NVIDIA DLSS 4.5 Ray Reconstruction Review
NVIDIA DLSS 4.5 Ray Reconstruction boosts image quality with a new AI-powered model that uses 20% more parameters and 35% more compute, all without sacrificing performance. Our testing shows that ray-traced visuals are enhanced, offering crisper, richer textures, especially in path tracing.
Garmin Fenix 9 and Fenix 9 Pro: Where to buy, release date, launch price
Go brighter, thinner, and get new stamina tracking from Garmin with these new models.
Apple Quietly Refreshes Magic Keyboards
Apple today quietly introduced updated Magic Keyboard models with revised U.S. English key labeling. On the new keyboards, the tab, caps lock, shift, return, and delete keycaps are labeled with glyphs rather than edge text. On the full-size models with a numeric keypad, the change goes further, with the home, page up, page down, end, clear, and enter keys also swapping their text labels for symbols.
Apple's latest Mac Mini runs on a new M6 chip, and starts at $899
The base model of the Mac Mini comes with 256GB of storage, 16GB of RAM, and costs $899.
Peacemaking Is Power Politics Again
The post-Cold War model of institutionalized peacemaking has waned. But influential capitals are still looking to end and avoid conflicts.
Apple Unveils New Mac Studio With M5 Max and M5 Ultra Chips
Apple today announced new Mac Studio models with the M5 Max and M5 Ultra chips, the latter of which is the company's most powerful chip ever. The M5 Ultra chip includes an up to 36-core CPU that contains 12 super cores and 24 performance cores. It delivers up to 1.25x higher single-threaded performance and up to 1.3x higher multithreaded performance than the M3 Ultra from the previous generation Mac Studio.
Apple announces new Mac Mini and Mac Studio models with AI upgrades
"With these frameworks and new chips, developers can run and fine-tune large AI models locally on their Mac," Apple said.
ASUS ROG adds three 24.5-inch OLED esports monitors up to 720Hz, new Fast IPS monitor reaches 920Hz
ASUS has confirmed three 24.5-inch OLED esports monitors at Gamescom, including the 720Hz ROG Swift PG259QWS Ace. Two Fast IPS models also offer selectable modes up to 920Hz.
Apple Watch Series 12 to Get Fitness Tracking Upgrades
Apple's new Apple Watch models will feature upgrades to fitness tracking, according to Bloomberg's Mark Gurman, who otherwise characterized this year's changes to the lineup as "minor". Apple's new Apple Watch models will feature upgrades to fitness tracking, according to Bloomberg's Mark Gurman, who otherwise characterized this year's changes to the lineup as "minor".
Introducing: A New Generation Of The Omega Speedmaster 38 MM
Seven new models bid farewell to oval subdials on the path to more classic Moonwatch styling.
New iPad Mini With Four Upgrades Expected to Launch by Late October
Apple is planning to release a new iPad mini model with an OLED display "by the end of October" this year, according to Bloomberg's Mark Gurman. Apple already updated the iPad Pro with an OLED display in May 2024, but the iPad mini, iPad Air, and entry-level iPad are all still equipped with LCD screens. The move to OLED technology would result in the next iPad mini offering improved image quality, thanks to richer colors and higher contrast ratio with true blacks.
AirPods 5 to Launch as Early as Next Month
Apple is planning to announce two new AirPods 5 models as early as Wednesday, September 9, according to Bloomberg's Mark Gurman. Like the AirPods 4, the AirPods 5 will be available with and without active noise cancellation, he said. AirPods 5 references were already discovered in macOS 26.7 code last week. AirPods 5 references were already discovered in macOS 26.7 code last week.
Apple could debut two AirPods 5 models next month, per report
Although AirPods with cameras aren’t expected until next year, Bloomberg reports that Apple could announce the AirPods 5 as early as next month’s expected iPhone event.
Garmin Fenix 9 (Pro) leaks in pictures, including AMOLED and solar MiP versions
A leak is said to show the design of the Garmin Fenix 9 and the Garmin Fenix 9 Pro for the first time, with the top-of-the-line model featuring both an AMOLED version and a MiP variant with integrated solar cells. Although Garmin is revamping the design, the smartwatches remain true to the Fenix look.
Climate Model Benchmarking: Building Trust and Advancing Science
A critical look at how well climate models work, how to successfully use them, and the anticipated challenges for their next-generation development.
One Of Our Most Detailed Views Yet Of Betelgeuse Reveals Unexpected Long-Lasting Structures In Its Inner Atmosphere
There are hot spots that have been there for longer than models predicted.
Garmin's next big Apple Watch Ultra rival leaks in 18 different models ahead of this week's launch
An extensive Garmin Fenix 9 family of upgraded multisport rugged smartwatches is just about guaranteed to see daylight on August 25.
Alibaba launches Wan3.0 AI video model after $10 billion share sale
Alibaba officially rolled out its latest AI video generation-model Wan3.0 on Monday with enhanced capabilities after the Chinese internet giant launched a $10 billion share placement to fund rising AI spending.
Microsoft’s Skala AI Model Joins CP2K to Improve Quantum Mechanical Simulations
The Skala model is now available, giving researchers an AI-based exchange-correlation functional for quantum mechanical simulations.
This Week On FloRacing: Aug 24
See the FloRacing schedule for Aug. 24-30, featuring High Limit Racing, USAC, NASCAR, late models and more.
Pixel 11's new HiLight feature makes its flashlight noticeably dimmer than prior models
A brightness comparison shows that the Pixel 11 Pro's camera flash and torch are dimmer than those of rivals like the Galaxy Z TriFold and iPhone 17e. The likely culprit is the same diffuser that powers the HiLight notification ring, which sits directly over the main flash LED.
Another major Android phone maker caves to soaring RAM costs with price hikes
Xiaomi has announced that it will start hiking smartphone prices across various models and configurations starting September.
10 Reliable Instant Photo Printers Worth Buying In 2026
Print your favorite photos anywhere with these reliable instant photo printers, featuring models from Canon, Fujifilm, Polaroid, Kodak, and more.