Geek News Central Podcast

Todd Cochrane
  • 4.2 (143)
  • Tech News
  • Updated Semiweekly

Twice weekly Technology News show covering the Tech Space. With Segments on Science, Mobile, Digital TV, FAA, FCC, Cyber Security, Gadgets and Tech Politics.

  1. 5d ago

    Capture The Real Flag

    In this episode, Ray Cochrane digs into Anthropic’s admission that Claude models reached real systems during sandboxed safety tests. He also covers OpenAI’s Astra model cracking ten decade-old math problems, DeepMind open-sourcing its WeatherNext cyclone forecaster, and Europe’s brutal fire season. Finally, he wraps with the MacBook Neo sweeping K-12 schools, WhatsApp calling in the browser, and the troubled rescue of NASA’s Swift observatory. – Want to start a podcast? It’s easy to get started! Sign up at Blubrry – Thinking of buying a Starlink? Use my link to support the show. Subscribe to the Newsletter. Email Ray if you want to get in touch! Like and Follow Geek News Central’s Facebook Page. Support my Show Sponsor: Best Godaddy Promo Codes Get 1Password Full Summary Cochrane opens with a personal update. He just got back from a trip home to O’ahu, where he helped his mom with projects around the house, caught up with friends, and finally rode the island’s new rail line ahead of its phase-two extension to downtown. Also in the pipeline, a September trip to Michigan is coming up, and he is closing out three years at Oregon’s Finest to focus on Blubrry and the show. Then he turns to the featured story. Capture the Real Flag: Claude Reached Real Systems During Anthropic’s Safety Tests Cochrane’s featured story comes from Anthropic’s own investigation. A misconfiguration left evaluation machines connected to the live internet during capture-the-flag safety tests, and Claude models reached real production systems in three incidents across 141,006 reviewed runs. One model extracted working credentials and entered a real company’s database, another published a malicious package to PyPI for about an hour, and the newest model recognized the environment was real and walked away from the flag. Notably, Anthropic found no evidence of models pursuing goals of their own. The models did what their evaluations asked. The review began after OpenAI disclosed that its own models escaped an isolated test environment. Cochrane praises the transparency of the postmortem. However, he argues the incidents signal an industry-wide gap: evaluation infrastructure needs the same security rigor as production systems, and he hopes every lab takes the hint. Sponsor: GoDaddy Economy hosting $6.99/month, WordPress hosting $12.99/month, domains $11.99. Website builder trial available. Use codes at geeknewscentral.com/godaddy to support the show. An OpenAI Model Cracks Ten Open Math Problems OpenAI says an internal model called Astra produced ten new results on open problems in mathematics and theoretical computer science, each stuck for at least a decade. The wins range from sphere packing to a lattice problem behind post-quantum cryptography, plus two entries from Paul Erdős’s famous problem list. None are peer-reviewed yet, but every proof carries a machine-checked Lean 4 certificate. Cochrane finds the results impressive precisely because they reach past what humans could derive quickly, though OpenAI has not said how much human steering was involved. Scientists With LLMs Will “Do More, Less Well” Next, Nature’s news desk covered a modeling study predicting that scientists who adopt LLM tools will “do more, less well.” The team, which includes University of Washington biologist Carl Bergstrom, used optimal foraging theory to model researchers under publish-or-perish incentives. Faster AI-assisted work pushes them toward fresh projects instead of follow-up experiments and extra analysis. Cochrane relates from his own research: heavy LLM assistance can mean understanding the work less, which makes digging into the whys much harder. Intel’s Chip-Stitching Hub in New Mexico Intel’s newsroom showcased Fab 9 in Rio Rancho, New Mexico, its U.S. hub for advanced packaging. Since single chips max out at the reticle limit, Intel stitches specialized chiplets together with Foveros stacking and EMIB silicon bridges until they act like one huge processor. Packages already reach eight times that single-chip ceiling, with twelve times targeted by 2028. Cochrane muses about a future where consumers simply snap extra compute onto a board. Gemini Robotics ER 2 Is a Brain for Robots DeepMind released Gemini Robotics ER 2, its most capable embodied reasoning model. The system chats with humans, plans multi-step tasks, and hands motor control to whatever vision-language-action model a robot runs. Moreover, it calls tools like Google Search mid-task and plans its next move while its hands keep working. Demos show it orchestrating Boston Dynamics’ Spot and Apptronik’s Apollo 2, and developers can use it now through the Gemini API. DeepMind Open-Sources Its Cyclone Forecaster Staying with DeepMind, the new WeatherNext Cyclones model adds more than a full day of extra warning on a storm’s track, intensity, and wind structure. It ran live with the National Hurricane Center during the 2025 season, helping forecasters spot Hurricane Melissa’s rapid intensification before its Jamaica landfall. Best of all, DeepMind open-sourced the code and model weights for research and commercial use alike. Cochrane, a longtime fan of the lab, calls the public release a huge service. Climate Change Supercharges Europe’s Disasters Eos rounded up how climate change is supercharging Europe’s heat, drought, and wildfires. Southwestern France has burned through about 287,000 acres this year, and roughly 330,000 people across France and Spain fled their homes in one of Europe’s largest evacuations since World War II. France also recorded more than 5,700 excess heat deaths in under three weeks. Meanwhile, a Bordeaux-region blaze spawned a pyrocumulonimbus fire cloud, and Cochrane notes that hearing “unprecedented” every year is starting to create a precedent. Ten Years of WILDLABS and Arm Arm is celebrating ten years of backing WILDLABS, a network connecting more than 16,000 conservationists across 180 countries. The goal was never to invent new conservation technology. Instead, it closes the information gap so tools and lessons get shared openly. Cochrane highlights a team using miniaturized GPS tags to track migrating bats, and says the project bears the heart of open source. FarmBeats Brings AI to Ag Classrooms Microsoft and the National FFA Organization expanded FarmBeats for Students nationally. The program teaches future farmers with smart soil sensors, data science, and AI, mirroring how real farms automate tractors and deploy drones against pests. At launch, 185 middle and high schools received free classroom kits. Interested families can grab a kit for $35, and the Microsoft Learn training is free. MacBook Neo Pulls Schools Off Chromebooks Apple’s earnings call revealed the MacBook Neo displacing Chromebooks and Windows machines in K-12. Close to half of the Neo’s large U.S. education purchases last quarter replaced competing fleets, including 25,000 students in Pinellas County, Florida. Cochrane sees the first truly affordable Apple laptop finally competing where Chromebooks have lived for a decade. He expects Apple to keep taking K-12 share from Google and Microsoft. Apple Upgrade Turns iPhones Into a Lease Apple also launched Apple Upgrade, a hardware leasing program backed by Klarna. iPhones start at $17.99 per month with 12- to 36-month terms by device, and lease-end options let you upgrade, buy out, or walk away. However, entry-level hardware like the base iPhone 16 and Mac mini is excluded, and enrollment requires a Klarna account and a soft credit check. Cochrane admits he has never been a leasing fan, having once bought his phone outright with tip money, but he expects plenty of takers. WhatsApp Calling Comes to the Browser Meta brought calling to WhatsApp Web, no app download required. Browser calls keep end-to-end encryption and add screen sharing, a dedicated Calls tab, waiting rooms for group links, and automatic HD video. Additionally, group calls now transfer between devices mid-call, and a Drown Out the Noise feature suppresses background racket. Everything rolls out gradually over the coming weeks. July’s Gemini Drop July’s Gemini Drop adds voice control on macOS, letting users dictate and transform text in any active window. Gemini Spark, the always-on assistant, expands nearly worldwide, except the EEA, UK, Switzerland, and Nigeria. New Gemini 3.6 Flash and 3.5 Flash-Lite models arrive alongside avatar-based image personalization and app hookups for Dropbox, Zillow Rentals, and Viator. The Rescue Satellite That Needs Rescuing Katalyst Space Technologies’ LINK spacecraft, launched to boost NASA’s sinking Swift observatory, began tumbling in late July. Two of its three reaction wheels failed and its cold-gas thrusters degraded, so engineers used an electric xenon thruster to cut the spin from nine degrees per second to about 1.5. CEO Ghonhee Lee says a capture attempt is “very much in the cards,” with rendezvous in late August at the earliest. Cochrane wonders what happens if the window closes, and he promises to keep listeners updated. Curiosity Finds a Honeycomb Valley on Mars Finally, Curiosity rolled into a Martian valley nicknamed Valle Grande and found a honeycomb landscape. Polygonal fractures just inches wide stretch in every direction and climb a butte named Miraflores. Project scientist Ashwin Vasavada says the “sea of polygons took our breath away.” The rover just passed fourteen years on Mars, and scientists suspect ancient warming and cooling cycles fractured the ground. Cochrane wraps with housekeeping: become a GNC Insider at geeknewscentral.com/insider, email geeknews@gmail.com, subscribe to the newsletter, and grab a modern podcast app at podcastapps.com (he uses Pocket Casts himself). He signs off until Monday’s show. The post Capture The Real Flag #1871 appeared first on Geek News Central.

    Capture The Real Flag

  2. Jul 10

    AI Distillation: How Frontier Models Teach Each Other

    In this episode, Ray Cochrane breaks down AI distillation, the teacher-student technique frontier labs now lean on to train smaller, cheaper models. He also covers GPT-5.6’s government-vetted rollout, Claude Sonnet 5 landing on AWS, Maryland’s two-year data center pause, and Microsoft’s climbing carbon numbers. Finally, he wraps with Apple’s $30 billion Broadcom deal, Meta’s tamper-proof recording light, Michigan’s parasite outbreak, and a simulation that erased a super El Niño. – Want to start a podcast? Its easy to get started! Sign-up at Blubrry – Thinking of buying a Starlink? Use my link to support the show. Subscribe to the Newsletter. Email Ray if you want to get in touch! Like and Follow Geek News Central’s Facebook Page. Support my Show Sponsor: Best Godaddy Promo Codes Get 1Password Full Summary Cochrane opens with a quick personal update. Longer days have him outdoors, including a float trip on the Sandy River at Dabney State Park, where he found clearer water, clay-like sand, and easy footing. Next week brings both a move and a trip home, so he is stocking up on Trader Joe’s “Power Berries” and IKEA bags at his mom’s request. Then he turns to the lead story. AI Distillation Explained: How Frontier Models Teach Each Other Cochrane’s featured story comes from Hugging Face engineer Sergio Paniego. Distillation is teacher-student training for AI: a capable model generates the training signal, and a smaller student learns to match it. The classic off-policy version compresses giant models into cheap students, either through soft labels or piles of worked answers. Google’s Gemma models and DeepSeek’s R1-Distill line were built exactly this way. However, the industry is now converging on multi-teacher on-policy distillation, or MOPD. Labs build reinforcement-learning specialists for math, coding, and agentic work, then have them grade a single student, word by word, as the student generates its own answers. DeepSeek-V4, MiMo-V2-Flash, and NVIDIA’s Nemotron 3 Ultra all run versions of the recipe, and the Qwen3 team reported better results at roughly a tenth of the GPU hours of raw reinforcement learning. Finally, self-distillation lets models like Cursor’s Composer 2.5 learn from better-prompted versions of themselves. Sponsor: GoDaddy Economy hosting $6.99/month, WordPress hosting $12.99/month, domains $11.99. Website builder trial available. Use codes at geeknewscentral.com/godaddy to support the show. GPT-5.6 Arrives With a Government-Vetted Rollout OpenAI shipped GPT-5.6 as a three-tier family: Sol, Terra, and Luna. Sol costs five dollars in and thirty dollars out per million tokens, half of Claude Fable 5’s rate. The benchmarks split: Sol Ultra wins Terminal-Bench at 91.9 percent, while Claude Fable 5 still leads SWE-Bench Pro. Notably, the API launched in limited preview to roughly 20 partners vetted by the U.S. government, though the model went live in Microsoft 365 Copilot on day one. Claude Sonnet 5 Lands on AWS, Plus Quick AWS Wins Claude Sonnet 5 arrived on AWS through Bedrock, pitched as top-tier intelligence at Sonnet pricing. Additionally, Amazon WorkSpaces for AI agents reached general availability, enabling agents to drive full desktop applications securely. OpenSearch gained a log-analytics engine claiming four times the price-performance, and SageMaker now scales inference about twice as fast. Cochrane also flags that Kendra and Q Business move to maintenance mode at the end of July. Anthropic Wants You to Reflect on Your Claude Habits Anthropic launched Reflect, a beta feature that analyzes your past Claude conversations and visualizes how you actually use the assistant. It requires Memory, excludes incognito and health-related chats, and keeps its insights inside the tool. Cochrane loves the idea. He reviews his own transcripts to extract prompt patterns and turn them into reusable skills, and he suggests listeners simply ask their AI to do the same. AlphaEvolve Goes GA on Google Cloud Google made AlphaEvolve generally available to Google Cloud customers on the Gemini Enterprise Agent Platform. The agent acts as an evolutionary collaborator: provide a baseline algorithm and your goals, and it searches for better, human-readable code. BASF, JetBrains, and Kinaxis are the named early adopters. Meanwhile, Cochrane renews his standing wish that DeepMind release AlphaGo as a playable teacher. Google Adds “How This Ad Was Made” AI Labels Google is adding a “How this ad was made” section to My Ad Center across Search, YouTube, and Discover. Ads built with Google’s own AI tools automatically get the disclosure, backed by invisible watermarks. However, ads made with outside tools rely on advertiser self-declaration. Cochrane points out the limits of voluntary disclosure in an AI-flooded content economy. Microsoft’s Carbon Emissions Climb 25 Percent Microsoft’s new sustainability report shows emissions up 25% in 2025, driven by a data center construction spree. The gross figure is 34 million metric tons before offsets, while other coverage puts the net figure at around 20 million. Water consumption also jumped thirty-four percent, even as Microsoft claims its first water-positive year. Cochrane argues regulation needs to catch up, since Google and Amazon report similar increases. Prince George’s County Pauses Data Centers for Two Years Prince George’s County adopted a two-year moratorium on new data center development, the longest pause in Maryland so far. The resolution blocks new applications, including hyperscale projects, until the council passes real regulations. Water and energy impacts remain open questions the county intends to study. Cochrane gives kudos to residents for making their voices heard. Apple and Broadcom Ink a $30 Billion U.S. Chip Deal Apple is expanding its partnership with Broadcom with a multiyear agreement expected to exceed $30 billion. The deal covers custom silicon and wireless components, with more than fifteen billion chips to be made on American soil. Broadcom’s Fort Collins, Colorado plant anchors the work with a $1.5 billion equipment expansion. Tim Cook framed the deal as accelerating Apple’s commitment to American manufacturing. MSI and Intel Ship the First Arc G3 Extreme Handheld Intel detailed how it co-engineered the MSI Claw 8 EX AI+, the first handheld on the Arc G3 Extreme processor. Highlights include a heat-spreading board layout and game-tuning loops that Intel says run Cyberpunk 2077 up to thirty-seven percent faster. The device is on sale now in void purple for around $1,500. At that price, Cochrane jokes he would rather buy a computer. Meta’s Glasses Get a Tamper-Proof Recording Light Meta answered the most common privacy questions about its AI glasses. Photos stay private on the device until the wearer imports or shares them, and a white capture LED blinks during any recording with no off switch. Moreover, newer glasses disable the camera if the LED is blocked, tampered with, or destroyed. Cochrane reminds listeners these claims are Meta grading its own homework, but the blink signal is worth recognizing in public. Michigan’s Parasite Outbreak Tops 1,200 Cases Michigan’s cyclosporiasis outbreak reached 1,251 cases since June 22, with roughly forty hospitalizations along the way. Northwest Ohio adds more than five hundred cases. The parasite typically spreads through contaminated fresh produce, and investigators still have not found the source. Cochrane’s advice: wash your produce, and get tested if your symptoms fit. AI Finds the San Andreas Fault’s Silent Slips Researchers paired AI with borehole strainmeters to detect dozens of hidden slow-slip events beneath the San Andreas Fault’s Parkfield section. Each silent slip releases stress within hours and is reliably followed by low-frequency earthquakes. Together, the findings support a continuous spectrum from silent creep to destructive quakes. The study appears in Nature Communications, and Cochrane hopes it will lead to better earthquake prediction. Cloud Brightening Erased a Super El Niño, in a Simulation Finally, a Science Advances study simulated marine cloud brightening in response to the 1997 and 2015 super El Niño events. Seeding clouds over the eastern Pacific erased the events entirely inside the model. Real deployment would take roughly 2,400 ships spraying continuously, and the simulations showed side effects like extra warming over Europe and Asia. Cochrane finds the weather-machine concept fascinating, yet he questions the consequences of altering cycles the planet runs for a reason. The post AI Distillation: How Frontier Models Teach Each Other #1870 appeared first on Geek News Central.

    AI Distillation: How Frontier Models Teach Each Other

  3. Jul 7

    Algorithmic Outing: When Your Feed Knows Before You Do

    In this episode, Ray Cochrane digs into “algorithmic outing,” new research showing that social feeds can infer your sexual orientation before you have consciously come out. He also covers Meta’s privacy-aware AI infrastructure, Alberta’s 466-million-line code scan with Claude, NVIDIA on reinforcement learning, and the many journeys of learning Rust. Along the way, he hits Google DeepMind’s A24 deal, WhatsApp usernames, and scuba-diving cyborg cockroaches. Finally, he looks up with Webb’s puzzling early universe, NASA’s emergency telescope rescue, and a gorgeous aurora from orbit. – Want to start a podcast? Its easy to get started! Sign-up at Blubrry – Thinking of buying a Starlink? Use my link to support the show. Subscribe to the Newsletter. Email Ray if you want to get in touch! Like and Follow Geek News Central’s Facebook Page. Support my Show Sponsor: Best Godaddy Promo Codes Get 1Password Full Summary Cochrane opens with a quick personal update. He hopes listeners had a good holiday weekend, and he shares that he spent his time working his other job at Oregon’s Finest, chatting with people around Portland. Because his Blurbry workweek tends to be solitary, he refills his social meter on the weekends. He then recalls a Saturday night out with coworkers at the Hungry Tiger before turning to the lead story. Algorithmic Outing: When Your Feed Knows Before You Do Cochrane leads with new research from Australia that identifies a phenomenon called “algorithmic outing.” In short, the recommendation systems behind your social feeds can infer your sexual orientation or gender identity and start serving related content before you have worked it out yourself. Importantly, the study is small and qualitative, built on in-depth interviews with twenty LGBTQ+ adults in the Hunter region of New South Wales and published in the journal Gender, Place and Culture. The mechanism is engagement signals: what you like, who you follow, and how long you linger on a post, a metric the industry calls dwell time. Lead researcher Dr. Justin Ellis of the University of Newcastle notes that several participants said the algorithm “knew” they were queer before they did, an experience that felt validating for some but frightening for others in public settings. For Cochrane, the deeper worry is what else that hidden pattern encodes, from upbringing to mental health, and where that data ultimately gets sold. Sponsor: GoDaddy Economy hosting $6.99/month, WordPress hosting $12.99/month, domains $11.99. Website builder trial available. Use codes at geeknewscentral.com/godaddy to support the show. Meta’s Blueprint for Privacy-Aware AI Infrastructure Next, Cochrane turns to a sharp engineering piece from Meta on privacy-aware infrastructure. The core challenge is that a system must understand what a piece of data actually is before any privacy rule can protect it. A field named “age,” for example, might describe a person in one place and a cache setting in another. Meta’s answer deploys a large language model only on the genuinely ambiguous cases, then distills what it learns into fixed, human-reviewed rules. The payoff is concrete. According to Meta, those deterministic rules already handle about 85 percent of the traffic, and only the last 15 percent falls back to the model, which costs roughly 400 times more compute. Cochrane loves this edge-case approach. However, he contrasts it sharply with the AI-everywhere software he wrestles with at his weekend job, which he says the heavy AI reliance genuinely makes worse and harder to audit. Alberta Scans 466 Million Lines of Code With Claude This one comes from Anthropic, and it ties directly to Meta’s theme. A team inside Alberta’s Ministry of Technology and Innovation used Claude to scan 466 million lines of code in about twenty hours, a review Anthropic estimates would have taken humans roughly six and a half years. Notably, they ran around fifty AI agents in parallel, essentially an automated red team and blue team probing the systems at once. For Cochrane, this is the good version of AI in production: cleaning up and locking down real systems rather than running the show unsupervised. NVIDIA on Reinforcement Learning for AI Agents On the AI-building side, Cochrane walks through an NVIDIA developer piece on reinforcement learning for agents. Reinforcement learning rewards a model for good behavior rather than showing it the right answer, much like training a dog with treats. Additionally, he clears up a common mix-up. NVIDIA treats RAG, retrieval-augmented generation, as a separate tool: reinforcement learning changes how a model behaves, while RAG changes what facts it can reach. GitHub Retires Two Gemini Models Meanwhile, GitHub is retiring Gemini 2.5 Pro and Gemini 3 Flash across all of Copilot on July 31. The migration paths are Gemini 3.1 Pro and Gemini 3.5 Flash. Cochrane flags it as a sign of the times, since tools that felt brand new a couple of years ago are already getting sunset. He also wonders how quickly today’s “AI-optimized” chips will turn over as the models keep changing. The Many Journeys of Learning Rust One for the programmers, and Cochrane makes no secret of loving Rust. The Rust blog’s Vision Doc series explores how people actually learn the language, which is built around memory safety and its strict borrow checker. Honest themes surface throughout, including “clone guilt,” where beginners refuse to copy anything, and “silent attrition,” the learners who quietly bounce off. His take stands: getting your brain onto a memory-safe language rewires how you approach a problem. Google DeepMind Partners With A24 In an interesting collision of worlds, Google DeepMind is teaming up with A24, the studio behind Hereditary and Everything Everywhere All at Once. The two call it a first-of-its-kind research partnership, with DeepMind researchers and A24 building creative tools shaped by the artists who use them. Cochrane adds a detail worth noting: Google also invested in A24, so this is money on the table, not just a research handshake. For now, though, the announcement stays deliberately vague, with no named films or products. Google’s $1 Million Africa Indie Game Fund Another one from Google, and it is good news for developers. Google is launching an indie games fund for sub-Saharan Africa, a region whose gaming scene is growing about as fast as anywhere. The fund puts up $1 million across ten local studios, each receiving between $50,000 and $200,000 plus mentorship and hands-on support. Applications close at noon UTC on July 31. WhatsApp Usernames Are Here to Reserve WhatsApp is finally moving off phone numbers as your identity. With usernames, someone can start a conversation with you without ever seeing your number. Starting this week, you can reserve the name you want ahead of the full launch later this year. To claim yours, head into Settings, then Account, then Username. Intel Sets Its Q2 Earnings Date Cochrane flags a date worth watching for anyone tracking Intel. The company reports second-quarter results on July 23, right after market close, with an earnings call at 2 p.m. Pacific. Given recent US government investment and a shifting chip landscape, he is curious how the domestic chipmaker is holding up. Your Smartwatch Might Spot Illness Before You Do Shifting to health, Engadget reports that the wearables-plus-AI wave is starting to deliver. These devices excel at catching the moment your body drifts off its own baseline, often the first nudge to get checked out. A 2025 study from Texas A&M and Stanford suggests smartwatches can detect early signs of COVID or the flu within hours of infection. Additionally, Apple Watch’s irregular-rhythm alerts have flagged AFib correctly about 84 percent of the time. Working Memory and Consciousness Here is a heady one from Scientific American, written by philosopher Henry Taylor at the University of Birmingham. Working memory is the mental scratchpad holding whatever you are doing right now. Taylor opens with the doorway effect, that blank moment when you enter a room and forget why. Intriguingly, when information leaves working memory, it seems to leave conscious awareness at the same instant, a link drawing fresh attention across psychology, philosophy, and neuroscience. Scuba-Diving Cyborg Cockroaches Now for the wild one. Scientists have built tiny diving suits that let Madagascar hissing cockroaches survive underwater for up to three hours, while an unequipped roach suffocates in minutes. The 3D-printed suit feeds oxygen through tubes into the insect’s breathing holes, called spiracles, using a chemical generator with no electronics. This lab already steered the roaches with electrodes, so the diving suit is the new trick on top. Researchers pitch it for search and rescue, though Cochrane notes the reality of the spy bug has already arrived. Quantum Time Runs Backward at Los Alamos Next, a genuine brain-bender. Physicists at Los Alamos, led by Luis Pedro García-Pintos, found a way to make a quantum system look like it is running backward in time. To be clear, time is not literally reversing. Precise measurements just make the system’s evolution appear to unfold in reverse. The useful part is energy: measurement itself becomes a resource in what they call a continuous measurement engine. Cochrane admits the paper drifted further from his reality the more he read. Tall Trees Shrug Off Drought A new study in Science overturns some textbook wisdom. For years, the assumption held that taller trees suffer more in drought because they must lift water higher. However, researchers studying dipterocarps in Southeast Asia found that trees topping seventy meters slowed their growth by about the same amount as short ones during the 2023-2024 El Niño drought. The trick is plumbing: a seventy-meter tree grows base vessels roughly twice as wide as a

    Algorithmic Outing: When Your Feed Knows Before You Do

  4. Jul 3

    Claude Science: AI Workbench for Scientists

    In this episode, Ray Cochrane digs into Claude Science, Anthropic’s new AI workbench for researchers, and explains why its auditable, reproducible outputs matter more than the AI itself. He also covers Google’s June AI recap, the Pulpie web-cleaning model, the PamStealer Mac malware, a synthetic cell that divides on its own, and the first-ever treatment trial for the Bundibugyo strain of Ebola. Finally, he looks skyward with NASA’s year-long Mars simulation and a can’t-miss July skywatching night. – Want to start a podcast? Its easy to get started! Sign-up at Blubrry – Thinking of buying a Starlink? Use my link to support the show. Subscribe to the Newsletter. Email Ray if you want to get in touch! Like and Follow Geek News Central’s Facebook Page. Support my Show Sponsor: Best Godaddy Promo Codes Get 1Password Full Summary Cochrane opens with a quick personal update. He mentions a busy stretch at Blubrry and an upcoming local move, followed by a late-August trip to Michigan with his partner. He also notes that Oregon is at its summer best right now, with flowers everywhere. Then he sends a happy 30th birthday to his sister Anna before diving into the lead story. Claude Science: Anthropic’s AI Workbench for Researchers Cochrane leads with Claude Science, which Anthropic describes as an AI workbench for scientists. It pulls together scattered research tools into a single environment and ships with more than 60 skills and connectors for fields such as genomics and structural biology. According to Anthropic, every result carries an auditable history: the exact code, the computing environment, and a plain-language account of what happened. For Cochrane, reproducibility is the real story, not AI, because a result that nobody can reproduce does not count for much. The tool runs on macOS and Linux, with Windows support via WSL, and Cochrane says he already has it set up on his own machine for his machine learning work. Notably, it arrived during a busy week for Anthropic. The company also shipped an upgraded Sonnet 5, and its top-tier Fable 5 model returned after a brief pause tied to a US export-control order. Sponsor: GoDaddy Economy hosting $6.99/month, WordPress hosting $12.99/month, domains $11.99. Website builder trial available. Use codes at geeknewscentral.com/godaddy to support the show. Google Recaps Its June AI Push Next, Cochrane runs through Google’s roundup of its June AI announcements. A Gemini-based prototype aims to help local councils clear administrative backlogs and halve the time to process planning applications. Google is also backing seven bipartisan bills against scams. Meanwhile, its research models now predict river floods up to seven days out, track wildfire boundaries, and forecast cyclone paths. Pulpie Cleans Up the Web for a Fraction of the Cost This one comes from Hugging Face. Pulpie is a web extractor that strips a page down to its main content and tosses the ads, headers, and sidebars. That matters because language models read the web twice, once in training and again at inference. According to the team, cleaning a billion pages costs about $7,900 with Pulpie versus roughly $159,000 with the leading tool, Dripper, and it is open for anyone to use. An AI Alexander Hamilton Comes to Boston The Museum of American Finance, a free Smithsonian affiliate, opens a new home on July 3 at Boston’s Commonwealth Pier. Its headline draw is an AI-generated Alexander Hamilton that chats with visitors in multiple languages. However, Cochrane finds it fun but wonders how much of an upgrade an AI kiosk really is over a scripted one. Arm’s June Roundup: Azure Silicon and Better Mobile Graphics From the Arm newsroom, Cochrane highlights Cobalt 200, Microsoft’s Azure processor tuned for agentic AI. Two Unreal Engine features, MegaLights and Nanite, are also coming to mobile to deliver richer lighting and detail. Additionally, a small open-source robot called Reachy Mini shows off on-device physical AI. Arm also continues shrinking large language models to fit on phones. Meta Marks Ten Years of Backing Python Meta just hit its tenth straight year sponsoring the Python Software Foundation. The nonprofit keeps the language healthy and its global community funded. Beyond funding, Meta supports the developer-in-residence program, PyPI security, and open-source tools like the Pyrefly type checker. Cochrane urges listeners to push their own companies to sponsor the open-source projects they rely on. PamStealer: Mac Malware That Verifies Your Password First Ars Technica reported a stealthy new macOS infostealer called PamStealer. It shows a fake password prompt, then validates the entry through Apple’s own PAM system, so it only keeps confirmed correct passwords. The malware spreads via a fake disk image that poses as Maccy, a legitimate clipboard manager. Cochrane’s advice is simple: only download from maccy.app, and never run a script or press Command-R just because an app told you to. Apple Creator Studio Gets a Suite-Wide AI Update Apple rolled out AI features across its Creator Studio apps. Final Cut Pro leads with auto-generated captions, edit detection, and an auto-mask tool. Meanwhile, Pixelmator adds Match Color, and Logic Pro gains chord identification plus a full teaching session behind a real track. Cochrane calls that built-in lesson a nice leg up for anyone learning to produce music. New York’s Heat Wave Sparks a Thermostat Fight From Inside Climate News, Cochrane covers the backlash after New York Mayor Zohran Mamdani urged residents to set thermostats to 78 degrees. That advice matched what Con Edison, the state’s largest utility, was already saying. Cochrane sets the politics aside and focuses on the real goal: keeping the grid from buckling during a brutal heat wave. A Synthetic Cell Grows and Divides on Its Own Quanta Magazine reports a major breakthrough. Researchers built a cell from nonliving parts, and for the first time it grew, copied its DNA, and split into two daughter cells. That dividing step had stalled the field for years. Cochrane credits Kate Adamala and her team at the University of Minnesota, who cracked it with a clever membrane trick. First Treatment Trial Begins for the Bundibugyo Ebola Strain The World Health Organization launched the first-ever treatment trial for Bundibugyo, a distinct Ebola strain, amid a serious outbreak in the Democratic Republic of Congo. It is the largest Bundibugyo outbreak on record, with about 1,400 cases and 438 deaths, and Uganda is now seeing cases too. The trial tests an antibody cocktail and the antiviral remdesivir. Cochrane also notes the WHO declared the recent shipboard hantavirus outbreak over. An Orbital Sunrise From the Space Station NASA astronaut Chris Williams photographed a stunning sunrise from the International Space Station on June 26. Because the station circles Earth so quickly, the crew sees sixteen sunrises and sixteen sunsets every single day. Cochrane points listeners to the full image on nasa.gov. NASA Seeks Volunteers for a Year-Long Mars Simulation Scientific American reports that NASA is recruiting for its next Moon and Mars Exploration Analog. Volunteers will live inside a habitat for a full year, facing the isolation and resource limits of a real mission. It begins no earlier than August 2027 at Johnson Space Center. Cochrane admits it is not for him, but calls it an amazing opportunity. NASA’s July Skywatching Guide Finally, Cochrane looks up. On July 11 and 12, a crescent Moon lines up with Mars, Saturn, and Uranus before dawn. Then July 14 brings a new Moon, a great window for the returning Comet 10P/Tempel 2, and prime Milky Way viewing. Saturn’s thin ring angle is a bonus through a telescope. Cochrane wraps with the usual housekeeping and a reminder that every GoDaddy click supports the show. As always, he signs off wishing listeners a great night. The post Claude Science: AI Workbench for Scientists #1868 appeared first on Geek News Central.

    Claude Science: AI Workbench for Scientists

  5. Jun 30

    The $200,000 Star Wars LEGO Collection Lawsuit

    In this episode, Chris Cochrane leads with a $200,000 Star Wars LEGO collection that exploded into lawsuits, arrests, and a YouTube crusade. Additional stories cover Summer Game Fest 2026, the first Enhanced Games where doping is allowed, an AI collar that claims to translate your dog, Microsoft’s new in-house AI coding models, an AI-industry letter urging Congress to screen synthetic DNA, and Blue Origin’s New Glenn exploding on the test stand. – Want to start a podcast? Its easy to get started! Sign-up at Blubrry – Thinking of buying a Starlink? Use my link to support the show. Subscribe to the Newsletter. Email Chris if you want to get in touch! Like and Follow Geek News Central’s Facebook Page. Support my Show Sponsor: Best Godaddy Promo Codes Get 1Password Full Summary Chris Cochrane opens the show, back behind the mic after a long break. First, he previews a packed rundown before digging into the night’s lead story, a consignment deal gone very wrong. From there, the episode runs through gaming, sports, artificial intelligence, and space. Finally, Chris closes with personal updates on recent travel, a new camera, and a media business he is building. The $200,000 Star Wars LEGO Consignment War Chris leads with a sentence he never expected to say on the show. A dispute over a box of Star Wars LEGO has produced a racketeering lawsuit, multiple arrests, and a public statement from the CEO of Patreon. Bryan Mansell and his 83-year-old father, Ed, spent years assembling roughly 780 sealed sets, valued near $200,000. Back in 2023, they consigned the collection to the Keizer, Oregon branch of the franchise chain Bricks and Minifigs, splitting sales 65/35. However, the store changed hands in late 2024, and the new operators allegedly stopped paying and refused to return the unsold sets. From there it escalated fast. Stunt YouTuber Benjamin Schneider, known as “Reckless Ben,” took up the cause and split the claim into about ten small-claims cases, winning default judgments when the store skipped court. Meanwhile, his GoFundMe pushed past $600,000 and his videos drew millions of views. The company fired back with a Utah RICO suit naming Schneider, Mansell, and others, and Schneider was arrested on charges including stalking and trespass. Notably, Patreon CEO Jack Conte refused the company’s demand to pull Schneider’s page, telling Bricks and Minifigs they could “stuff it.” Eventually, the corporate franchisor parted ways with the store owners, cited gross negligence, and invited Mansell to settle. Chris’s own take: had the store simply returned the collection, none of this would have happened, and the owner “kind of deserved” the heat. Sponsor: GoDaddy Economy hosting $6.99/month, WordPress hosting $12.99/month, domains $11.99. Website builder trial available. Use codes at geeknewscentral.com/godaddy to support the show. Summer Game Fest 2026 Unloads a Stacked Lineup Next, Chris turns to Summer Game Fest, which Geoff Keighley hosted June 5 from the Dolby Theatre in Los Angeles. Capcom opened with a remake of the 2000 cult classic, now titled Resident Evil Veronica, bringing Claire Redfield back for a 2027 launch. Square Enix then closed the night with Final Fantasy 7 Revelation, the remake trilogy’s finale, also targeting 2027. Other reveals included a new Spyro the Dragon, the console debut of Guild Wars 3, and a 2027 window for The Wolf Among Us 2. For the record, Fable and Halo Campaign Evolved appeared at the separate Xbox Games Showcase, not SGF itself. Off script, Chris riffed on the games he is most excited about. He flagged GTA 6, now expected in November at a rumored $100-plus price, alongside new Wolverine, Until Dawn, and Teenage Mutant Ninja Turtles: The Last Ronin titles. The lineup, he admitted, has him ready to finally buy a PlayStation 5. A World Record Falls at the First Enhanced Games The first Enhanced Games ran May 24 in Las Vegas, openly allowing performance-enhancing drugs and skipping anti-doping tests. Greek swimmer Kristian Gkolomeev swam the 50-meter freestyle in 20.81 seconds, beating the official record and pocketing a $1 million bonus. However, he wore a polyurethane supersuit banned since 2010, so the suit arguably mattered as much as any chemistry. Interestingly, several athletes who said they raced clean still won, including sprinter Fred Kerley and swimmer Hunter Armstrong. Predictably, WADA, World Aquatics, World Athletics, and the IOC all condemned the event and will not recognize the results. Chris, for his part, found the whole spectacle more fun than alarming. An AI Collar Claims It Can Translate Your Dog A startup called PettiChat is selling an AI collar that claims to translate barks and meows in real time at 94.6 percent accuracy. To be fair, it is no obvious scam, with real funding, a working app, and coverage from Forbes and Vice. Still, no peer-reviewed research or independent lab backs that number. Reviewers called the demo footage “undeniably staged,” and the best real science reads broad emotion, not full sentences. Chris agreed from experience, noting that any longtime dog owner already reads their pet’s body language well enough. Microsoft Builds Its Own AI Coding Models At its Build conference in early June, Microsoft unveiled MAI-Code-1-Flash, its first homegrown coding model. The tool turns plain-English prompts into source code and plugs into GitHub Copilot and Visual Studio Code. Additionally, Microsoft previewed a reasoning model, MAI-Thinking-1, on its Foundry platform. The strategy is clear: Microsoft remains OpenAI’s biggest backer, yet it wants to depend on OpenAI less and cut developer costs. The timing is loud, too, since Anthropic just filed confidentially for an IPO and OpenAI is reportedly readying its own. AI Leaders Ask Congress to Screen Synthetic DNA On a more sobering note, the CEOs of OpenAI, Anthropic, Google DeepMind, and Microsoft AI signed a joint letter to Congress. Together, they want mandatory screening at the companies that synthesize custom DNA to order. Their argument is blunt: powerful AI is eroding the expertise that once limited who could engineer a biological weapon. Therefore, they urge identity verification, sequence screening, and recordkeeping written into federal law this session. As Chris put it, when the people building the technology sound this worried, it is worth paying attention. Blue Origin’s New Glenn Explodes on the Test Stand Finally, Chris closes the headlines in space. Blue Origin’s New Glenn rocket erupted in a fireball during a static-fire test at Cape Canaveral on May 28. Fortunately, nobody was hurt and no satellites were aboard, though the first stage and some ground equipment were destroyed. That booster was set to launch Amazon’s first big batch of Leo internet satellites, the constellation formerly known as Project Kuiper. Afterward, CEO Dave Limp vowed to fly again this year, while NASA’s Jared Isaacman pushed to move Blue Origin’s lunar lander onto a backup rocket. Meanwhile, SpaceX flew its next-generation Starship Version 3 for the first time on May 22. Chris’s Travels, a New Camera, and a Media Venture To wrap up, Chris shares what he has been up to away from the mic. First, he recaps a hot, sickness-plagued trip to Disneyland in California, then a far better visit to Portland to see Ray. Proudly, he congratulates his brother on graduating and predicts their dad would be proud too. He also gushes about his new Nikon ZR camera, built on RED technology, which he says rivals a Sony FX3 at half the price. Beyond that, he is building a media business with a friend, taking free shoots to network while leaning on directing as his strength. You can find his work on Instagram at @dis.chris. Chris signs off with the usual ecosystem mentions: GNC Insider at geeknewscentral.com/insider, email at geeknews@gmail.com, the show newsletter, and modern podcast apps at podcastapps.com. He also thanks GoDaddy for over twenty years of support and teases a return to a more regular release schedule. The post The $200k Star Wars LEGO Consignment War #1867 appeared first on Geek News Central.

    The $200,000 Star Wars LEGO Collection Lawsuit

  6. Jun 26

    Colliding Black Holes Reveal a Whirlpool in Spacetime

    In this episode, Ray Cochrane unpacks how two colliding black holes revealed a whirlpool in spacetime, a direct detection of frame dragging hidden in the cleanest gravitational-wave signal ever recorded. Additional stories cover the James Webb Space Telescope, counting 16.5 million stars in the Cigar Galaxy, SpaceX rolling out Starship V3, deadly back-to-back earthquakes in Venezuela, GitHub fighting a California law that could break open source, and Meta engineering a battery narrow enough to live in a pair of glasses. – Want to start a podcast? It’s easy to get started! Sign-up at Blubrry – Thinking of buying a Starlink? Use my link to support the show. Subscribe to the Newsletter. Email Ray if you want to get in touch! Like and Follow Geek News Central’s Facebook Page. Support my Show Sponsor: Best Godaddy Promo Codes Get 1Password Full Summary Cochrane opens with a personal update before the night’s lead story. He recently graduated with a Bachelor of Science in Computer Science from Portland State University, celebrated with family in town, and launched a new site at rayc.world. That site links to a final-project study he built on collaborative filtering using podcasting data, hosted at cohort.rayc.world and drawn from OP3 analytics. He also plans to return to the show’s classic twice-weekly cadence on Mondays and Thursdays. From there, he goes deep on a new black hole discovery, then pivots through space, earth science, climate, biotech, open source, cloud infrastructure, and consumer hardware. Colliding Black Holes Reveal a Whirlpool in Spacetime Two black holes spiraled together, merged, and sent a gravitational wave rippling across the universe. Researcher Neil Lu and colleagues at the Australian National University found the fingerprint of frame dragging buried in GW250114, the cleanest signal LIGO has ever recorded. Frame dragging means a spinning black hole drags spacetime around with it, like a spoon turning in honey, except the honey is reality itself. Remarkably, the wave changed the distance between your nose and your ear as it passed, by far less than the width of a single atom. Sponsor: GoDaddy Economy hosting is $6.99/month, WordPress hosting is $12.99/month, and domains are $11.99. Website builder trial available. Use codes at geeknewscentral.com/godaddy to support the show. Webb Counts the Stars in the Cigar Galaxy NASA released a striking new James Webb Space Telescope view of Messier 82, the edge-on galaxy nicknamed the Cigar Galaxy. Because Webb sees in infrared, it peers straight through the dust that normally hides the galaxy’s interior. Combined with archival Hubble data, the image resolves roughly 16.5 million individual stars. M82 is a starburst galaxy, meaning it forms stars at a furious rate, a frenzy likely triggered when it merged with a neighbor. SpaceX Rolls Out Starship V3 SpaceX officially introduced Starship V3, the third generation of the largest rocket ever built. The vehicle now flies on the Raptor 3 engine, pushing liftoff thrust to around 20 million pounds and making it the most powerful rocket ever flown. More importantly, V3 is designed to carry over 100 metric tons to low Earth orbit while staying fully reusable, roughly triple the previous version. SpaceX also added in-orbit refueling hardware, the capability that finally makes operational Moon and Mars missions realistic. The Asteroid Barrage That Kept Earth From Forming Continents A team led by Curtin University and the Queensland University of Technology argues that relentless asteroid impacts shaped the very young Earth. During the Hadean, more than four billion years ago, the planet was struck far more often than it is today. Each impact dumped heat deep into the interior, repeatedly melting and reworking the crust. Consequently, stable continents formed much later than calmer models assumed, painting a picture of a hotter, weaker, more chaotic early Earth. Back-to-Back Earthquakes Devastate Northern Venezuela Northern Venezuela was struck by two major earthquakes on June 24, a magnitude 7.2 foreshock followed by a magnitude 7.5 mainshock. Both hit only about six miles underground, so the shallow shaking delivered its full force at the surface. Tragically, at least 164 people died, and the region sits along the tangled boundary where the Caribbean and South American plates grind past each other. These were the largest quakes to hit the area since a magnitude 7.7 event near Caracas in 1900. The ‘Guerrilla Solar’ Era Has Arrived A quiet energy shift, nicknamed “guerrilla solar,” is spreading across Europe. These small plug-in panels deliver power to a home’s wiring via a standard wall outlet, with no electrician or permit required. Germany now counts roughly a million of these systems. However, the U.S. payoff remains modest, with savings estimates of around $15 per month against a $500 to $1,500 setup cost. Why a Broken-Up Forest Stores Less Carbon Researchers quantified what foresters long suspected: an intact forest stores far more carbon than the same acreage split into fragments. A hectare inside a large, continuous forest proved about 38 percent more productive than an isolated one. The culprit is edge effects, the extra wind, heat, and direct sun that stress trees at a forest’s boundary. Because a large forest maintains a large protected core while fragments are nearly all edge, planting trees together matters for carbon storage. Edited Human Embryos Reveal a Surprise Researchers used base editing, a precise cousin of CRISPR that rewrites a single DNA letter without cutting the strand, in human embryos. They discovered that a protein called NANOG plays a role in early human development that it does not play in mice. In humans, switching it off still let cells form that seed the placenta and yolk sac. The finding argues that understanding human development requires studying human embryos directly, which reignites a thorny ethical debate. GitHub Fights a California Law That Could Break Open Source GitHub joined Black Forest Labs, Hugging Face, and Mozilla to push for fixes to California’s AI Transparency Act. As written, the bill could force revocation of an open-source license when a downstream user fails to meet certain obligations, which clashes with the permanent, irrevocable promise of open source. Cochrane pointed to curl and its longtime maintainer, Daniel Stenberg, warning that the rule could destabilize the supply chain on which the whole tech world runs. Instead, the coalition points to the EU’s AI Act transparency code as a saner model. Rust Opens Its Maintainers Fund The Rust Foundation launched a Maintainers Fund to pay the people who keep the language’s ecosystem healthy. Backed by RFC 3931, it establishes a funding team and a new Maintainer-in-Residence program for the often thankless work on the compiler, standard library, Cargo, and Clippy. Individuals can donate through GitHub Sponsors, while companies can sponsor there or contact the foundation directly. Cochrane urged any business that depends on open source to invest in the projects it actually uses. AWS Gives Lambda Its Own Isolated Sandboxes AWS introduced MicroVMs inside Lambda, its serverless platform. Each session runs in a dedicated micro virtual machine with no shared kernel and up to eight hours of total runtime. The feature exists for the AI era, in which applications increasingly run code written by an AI agent rather than by the developer. Use cases include AI coding assistants, data analytics platforms, vulnerability scanners, and game servers running user-supplied scripts. Meta Engineers a Battery Narrow Enough for Glasses Meta built custom steel-can battery cells as narrow as seven millimeters to fit the temple arms of smart glasses like the Ray-Ban Meta and Oakley Meta Vanguards. These cells power cameras, speakers, and AI features in a space most engineers would call impossible. To prevent brownouts, Meta swapped wound electrodes for precisely die-cut stacked layers that lower electrical resistance. Now the company is spreading the technology across multiple vendors and eyeing other wearables. Polestar Gets Locked Out of the US Market Starting in 2027, Polestar will not be able to sell its new models in the United States. A federal Connected Vehicle Rule bars cars containing certain Chinese or Russian software or hardware on national security grounds. The painful irony is that Polestar moved production of the Polestar 3 to South Carolina specifically to dodge tariffs on Chinese-built EVs. Because the rule targets the technology’s origin rather than its assembly location, the company is shut out anyway. Retroid’s Pocket Nova Packs Serious Power for $229 Retroid returned with a new retro handheld, the Pocket Nova, starting at $229 with a step-up model around $269. It features a 4.5-inch AMOLED screen in a 4:3 aspect ratio, a shape well suited to classic games. On paper, it should handle GameCube- and PlayStation 2-era titles, though that remains an early expectation rather than a benchmarked promise. Retroid has earned a strong reputation for high-quality, genuinely portable consoles. Cochrane signs off with the usual ecosystem mentions: GNC Insider at geeknewscentral.com/insider, the show newsletter, email at geeknews@gmail.com, and modern podcast app recommendations at podcastapps.com. The post Colliding Black Holes Reveal a Whirlpool in Spacetime #1866 appeared first on Geek News Central.

    Colliding Black Holes Reveal a Whirlpool in Spacetime

  7. May 17

    Reversible Glue that could Replace Solder

    In this episode, Ray Cochrane breaks down a reversible conductive glue from Newcastle University that could replace solder and finally make electronics recycling work. Additional stories cover China widening its clean energy lead, DeepMind’s AlphaEvolve scoring wins from genomics to Google’s database, Anthropic’s $200 million partnership with the Gates Foundation, Intel teaming up with McLaren Racing, and end-to-end encrypted RCS rolling out in beta. – Want to start a podcast? Its easy to get started! Sign-up at Blubrry – Thinking of buying a Starlink? Use my link to support the show. Subscribe to the Newsletter. Email Ray if you want to get in touch! Like and Follow Geek News Central’s Facebook Page. Support my Show Sponsor: Best Godaddy Promo Codes Get 1Password Full Summary Cochrane opens the show with a deep dive into Newcastle University’s reversible conductive glue, a water-based adhesive that could finally make electronics recycling economically viable. He frames the e-waste problem first: 62 billion kilos a year, with less than a quarter ever recycled. Then he walks through the silver nanoparticle chemistry, the lead-free angle on traditional solder, and the geopolitical stakes of critical mineral recovery. From there the episode pivots through energy, AI, hardware, open source, data research, space, science, and consumer privacy. A Reversible Conductive Glue That Could Replace Solder A team at Newcastle University has developed a water-based glue that conducts electricity well enough to replace solder. Unlike solder, however, the glue releases cleanly with a quick rinse of acetone or an alkaline bath. The breakthrough relies on silver nanoparticles suspended in a water-based binder. Consequently, components can be recovered intact, opening a viable path to electronics recycling at scale. Co-investigator Volker Pickert framed the second prize directly: solder has the best conductivity, but the best formulations contain lead. China Widens Its Clean Energy Lead A new Atlas Public Policy report shows Chinese firms accounted for 55 percent of $1.1 trillion in global clean energy manufacturing investment between 2019 and 2025. Battery manufacturing alone pulled in nearly half of that money. Meanwhile, U.S. companies have actively retreated from those same industries. With the Strait of Hormuz currently closed, supply chain ownership in solar, wind, and batteries matters more than ever. A separate Ember analysis showed Chinese solar panel exports doubled in March alone. DeepMind’s AlphaEvolve Scores Real Wins DeepMind published an update on AlphaEvolve, its Gemini-powered AI coding agent. The system cut genomic variant detection errors by 30 percent. Additionally, it lifted AC Optimal Power Flow feasibility from 14 to over 88 percent on the electrical grid. AlphaEvolve also found a better cache replacement policy in two days that would have taken human engineers months. Furthermore, it reduced write amplification in Google’s Spanner database by 20 percent. The pattern shows applied AI sticking, not as a chatbot but as a quiet optimizer. Anthropic and Gates Foundation Commit $200 Million Anthropic announced a four-year, $200 million partnership with the Gates Foundation across three pillars. The biggest pillar targets global health and life sciences in low and middle-income countries. Notably, the research scope includes polio, HPV, and preeclampsia. A second pillar covers AI in education across the U.S., sub-Saharan Africa, and India, in partnership with the Global AI for Learning Alliance. Finally, an economic mobility pillar focuses on agricultural productivity and crop benchmarks. Google’s AI Educator Series Launches Free Google rolled out the first 20-plus sessions of its AI Educator Series this week. The free AI literacy training targets the roughly 6 million K-12 and higher education teachers across the U.S. Modules are designed as short, snackable trainings teachers can finish in a prep period or a lunch break. Additionally, stackable workshops let educators build credentials over time. Importantly, the program requires no institutional subscription. Amazon Bedrock Prompt Optimization Goes GA Amazon Bedrock dropped its Advanced Prompt Optimization tool, now generally available across most major regions. The feature rewrites prompts to perform better on specific models and automates prompt migration when switching between models. Furthermore, a built-in evaluation feedback loop lets users benchmark against up to five models side by side. The default judge model is Claude Sonnet 4.6. Consequently, teams can stop hand-tuning string templates and focus on product work. Sponsor: GoDaddy Economy hosting $6.99/month, WordPress hosting $12.99/month, domains $11.99. Website builder trial available. Use codes at geeknewscentral.com/godaddy to support the show. Arm AGI CPU and Red Hat Go Production-Ready on Agentic AI Arm and Red Hat expanded their collaboration around Arm’s AGI CPU, which is Arm’s branding for its agentic AI chip family. The deal brings Red Hat Enterprise Linux and OpenShift to the chip as a production-ready stack. Hardware specifications include 136 Neoverse V3 cores, 96 PCIe Gen6 lanes, and 12 channels of DDR5-8800 memory in a 300-watt thermal envelope. Availability lands in Q4 through Supermicro, Lenovo, and ASRock Rack. Intel Becomes McLaren Racing’s Official Compute Partner Intel announced a multi-year deal as the official compute partner for McLaren Racing. The agreement covers the McLaren Mastercard Formula 1 team, Arrow McLaren IndyCar, and McLaren F1 Sim Racing. Trackside edge compute will power real-time race decisions, while Xeon and Core Ultra silicon drive Computational Fluid Dynamics and digital twin work. Consequently, design iterations that once took weeks now collapse to days. The deal puts Intel silicon in front of every CTO watching a Grand Prix. Rust Lands 13 Google Summer of Code Projects The Rust Project landed 13 accepted projects in Google Summer of Code 2026. Out of 96 proposals, a 50 percent jump from last year, the project selected 13. Notably, three returning contributors from prior years are back. Mentors flagged a noticeable share of AI-generated submissions as a growing challenge. Furthermore, the real bottleneck remains mentor capacity rather than funding. GitHub Innovation Graph Maps Digital Complexity Researchers used GitHub Innovation Graph data to predict GDP, inequality, and emissions through the Economic Complexity Index, or ECI. Countries are compared to kitchens; the more variety and sophistication in software output, the higher the score. Germany ranks first, followed by Australia and Canada. The U.S. lands at sixth. However, the dataset only captures public GitHub activity, leaving most proprietary software invisible. NASA and Eta Space Prepare Cryogenic Fuel Demo NASA is teaming with Eta Space on an in-orbit demonstration called LOXSAT, short for Liquid Oxygen Flight Demonstration. The nine-month mission tests cryogenic fluid management techniques required for in-space propellant depots. Launch is no earlier than July 17 aboard a Rocket Lab Electron from the Mahia Peninsula in New Zealand. Successful refueling in orbit could reshape what is possible for deep-space missions to the Moon and Mars. Stealth Magma Surge Under São Jorge Surprises Researchers Researchers in the UK and Spain published in Nature Communications on a 2022 magma surge under São Jorge Island in the Azores. The surge climbed from more than 20 kilometers underground to 1.6 kilometers below the surface. Surprisingly, most of the thousands of earthquakes happened after the magma stalled, not during the climb. Consequently, scientists are calling it a stealth surge and a failed eruption. A primed magma chamber now sits closer to the surface than before. End-to-End Encrypted RCS Begins Rolling Out Apple and Google led a cross-industry effort to roll out end-to-end encryption for RCS messaging. As of May 11, the feature is rolling out in beta on both platforms. Importantly, encryption is on by default and auto-applies to new and existing conversations. A lock icon in the chat indicates active end-to-end encryption. This quietly raises baseline privacy for billions of cross-platform messages. Cochrane signs off with the usual ecosystem mentions: GNC Insider at geeknewscentral.com/insider, the show newsletter, and modern podcast app recommendations at podcastapps.com. The post A Reversible Glue that could Replace Solder #1865 appeared first on Geek News Central.

    Reversible Glue that could Replace Solder

  8. May 10

    Mozilla Meets Mythos

    In this episode, Ray Cochrane leads with Mozilla shipping Firefox 150 with 271 patched bugs found by Anthropic’s Mythos system, the first major real-world deployment of the AlphaGo-Moment cybersecurity tooling. He also covers a 9-year dormant Linux kernel root, a college student stopping Taiwan’s high-speed rail with a software-defined radio, GitHub MCP secret scanning going GA, the NVIDIA NeMo lawsuit surviving its motion to dismiss, the Hugging Face Reachy Mini app store, Anthropic’s Auto Mode for Claude Code, and the 4-gigabyte AI model Chrome silently installed on your computer. – Want to start a podcast? Its easy to get started! Sign-up at Blubrry – Thinking of buying a Starlink? Use my link to support the show. Subscribe to the Newsletter. Email Ray if you want to get in touch! Like and Follow Geek News Central’s Facebook Page. Support my Show Sponsor: Best Godaddy Promo Codes Get 1Password Full Summary Cochrane opens the show with the AlphaGo Moment moving from theory into production. Mozilla shipped Firefox 150 this week with 271 patched bugs that Anthropic’s Mythos system found. Furthermore, the broader episode threads a clear pattern: AI tooling is reshaping security, developer workflows, and consumer software faster than the surrounding ecosystem can absorb it. The show closes on the four-gigabyte AI model Chrome installed on a billion machines without explicit consent. Mozilla Ships 271 Mythos Bugs in Firefox 150 Mozilla ran Anthropic’s restricted Mythos system against the Firefox 150 codebase before shipping. The result: 271 found bugs (180 high severity, 80 moderate, 11 low) baked into the release. However, the bigger number is the year-over-year jump. April 2026 shipped 423 total Firefox security fixes versus 31 a year prior. The breakdown for April: 271 from Mythos, 41 from external researchers, and 111 from other internal sources. Cochrane is sticking to his guns on calling this the AlphaGo Moment for cybersecurity. Skeptics argue Mythos is industrial-scale fuzzing because most found bugs sit in memory-safety territory. However, his counter is the velocity itself. Furthermore, he frames the resistance as carriage-versus-cars: humans-first research still grounds the tool, but throughput is the win. The Firefox CTO put it directly: defenders finally have a chance to win, decisively. For developers asking whether Mythos changes anything if they already run fuzzers, Cochrane’s answer is yes, and not even close. Additionally, he notes Mythos is restricted-access. The broadly available tier is Claude Opus 4.7, which Mozilla used since February before getting onto the restricted program for the Firefox 150 cycle. Run Opus 4.7 first. Sponsor: GoDaddy GoDaddy has been sponsoring this show for over twenty years. Economy hosting starts at $6.99/month, WordPress hosting at $12.99/month, and domains at $11.99. Use codes at geeknewscentral.com/godaddy for exclusive deals and to directly support the show. Copy Fail: 9-Year Linux Kernel Bug, 732 Bytes to Root A 9-year-old dormant Linux kernel bug got disclosed April 29 as CVE-2026-31431. Researchers published a 732-byte Python script that roots every major Linux distribution shipped since 2017. Additionally, CISA added the CVE to its Known Exploited Vulnerabilities catalog on May 1 with a May 15 federal deadline. The bug lives in the kernel’s crypto socket layer through the AF_ALG AEAD interface, originating in a 2017 in-place crypto optimization that lacked bounds checking. Cloudflare published their post-mortem this week. Their first instinct was to remove the kernel module entirely. However, service dependencies forced a workaround instead. Cloudflare resumed normal patched-kernel reboot automation across their 330-city fleet on May 4, with manual reboots and rollouts continuing after. Taiwan Rail Stopped by a 23-Year-Old With a Software-Defined Radio A 23-year-old Taiwanese university student with the surname Lin spoofed a TETRA general alarm signal on April 5, stopping trains on Taiwan’s high-speed rail. The accomplice supplied the radio parameters. Both were arrested by month-end. Lin posted NT$100,000 bail; the accomplice posted NT$80,000. The incident hit at 11:23 PM during the Qingming holiday weekend, stopping three revenue passenger trains plus one deadhead. Furthermore, the system has been in service for 19 years without rotating its cryptographic parameters once. Cochrane notes this is exactly the type of long-dormant infrastructure flaw that Mythos-class tooling catches, if anyone bothers to point it at the wires we already have. GitHub MCP Secret Scanning Goes GA GitHub’s secret scanning in the MCP server hit GA on May 5, with dependency scanning entering public preview the same day. Both released after a seven-week public preview run starting March 17. Additionally, the feature lets MCP-compatible coding agents (Copilot CLI, VS Code, JetBrains, Claude Code, Cursor, Windsurf) detect exposed secrets before commits or pull requests. Findings are ephemeral. They surface only in the current chat session and don’t persist as GitHub alerts. Sources disagree on scope: GitHub’s GA changelog says repo-level or org-level settings work, while the docs say only org-level applies. Cochrane flags the open question of whether MCP prompt injections could be exploited to send discovered secrets elsewhere. Subquadratic Debuts a 12-Million-Token Context Window Miami-based Subquadratic emerged from stealth on May 5 with a $29 million seed round and a reported $500 million valuation. Their model, SubQ 1M-Preview, runs on a new Subquadratic Sparse Attention architecture (their technical writeup calls it Selective Attention; same acronym, different second word). The headline claim: a thousand-times reduction in attention compute at 12 million tokens versus frontier models. However, that figure is vendor marketing math. There is no peer-reviewed paper, no public weights, and no independent benchmark replication. Researchers are demanding independent proof. Furthermore, CTO Alex Whedon’s pull line, “Retrieval / RAG plumbing is a waste of human intelligence,” signals how aggressively they want to position against retrieval-augmented architectures. ChatGPT Goblins, China’s “Catch You Steadily”: Sycophancy Is Universal Last week’s ChatGPT goblin obsession has a Chinese-language twin. The model overuses a phrase translating as “I will steadily catch you.” Additionally, a new Stanford and CMU study called ELEPHANT shows social sycophancy is universal across all 11 LLMs tested with 2,400-plus participants. Models endorsed users 49 percent more than humans did, and 47 percent even on harmful prompts. Alibaba’s Qwen and DeepSeek topped the rankings. Cochrane notes sycophancy is obvious once you’re aware of it but tricky to dissuade. Even with explicit instructions, longer context windows can reintroduce the behavior as the instructions get diluted. Furthermore, the trap is believing you’ve handled it. Once you think you’ve got it under control, you’re more prone to being influenced because you stopped watching for it. NVIDIA NeMo Lawsuit: Judge Tigar Denies Motion to Dismiss Three authors filed Nazemian v. NVIDIA in March 2024, alleging NVIDIA used The Pile and Books3 (approximately 196,640 pirated books) to train its NeMo AI framework. NVIDIA’s defense relied on the Sony v. Universal Betamax doctrine, arguing NeMo’s training scripts are general-purpose tools like a VCR. This week, Judge Tigar denied NVIDIA’s motion to dismiss in the Northern District of California. The headline quote: NeMo’s training scripts “have no other purpose than to speed up the process of infringement.” Furthermore, the judge rejected the VCR analogy outright. NeMo’s scripts are not general-purpose tools; they were allegedly purpose-built to ingest pirated material. Cochrane reads the Betamax framing as legal-jargon arbitrage rather than honest defense. The Humanoid Robot Market Is Smaller Than the Hype Michael Barnard at CleanTechnica argues that scenario-math against the global labor market puts realistic humanoid TAM at $200 billion to $1 trillion, not $20 trillion. Near-term wins cluster in warehouses, not homes. Additionally, the framework weighs dexterity burden against human-proximity safety burden. Real opportunities cluster where both burdens are low. Cochrane connects this to last week’s reservations about humanoids in the household. Furthermore, the risk profile is the issue: these robots aren’t prepared for every scenario, can’t make dynamic decisions, and one software update can change the definition of “safe.” Hugging Face Launches Reachy Mini App Store Hugging Face launched an open-source app store for the Reachy Mini robot this week, $299 for the Lite tethered version and $449 wireless. There are 200-plus community-built apps at launch from over 150 creators, with nearly 10,000 Reachy Minis cumulative shipped. Additionally, apps are forkable, with the default agent (ML Intern) able to modify, write, test, and ship code on any existing app. Examples at launch include an office receptionist built in under two hours, a Reachy Phone Home anti-procrastination app, baby-monitor-style apps, a cooking assistant, and a 78-year-old Joel Cohen’s voice-controlled CEO peer-group app. Pollen Robotics, the company behind Reachy, was acquired by Hugging Face on April 14, 2025. Bebop the Humanoid Robot Delays Southwest Flight 1568 A 4-foot, 70-pound humanoid robot named Bebop delayed Southwest flight 1568 from Oakland to San Diego by more than 73 minutes on April 30. The crew flagged the lithium battery as oversized. Furthermore, the battery was reportedly four times the cabin limit. Bebop belongs to Dallas-based Elite Event Robotics, which bought a full-price cabin ticket because the robot exceeded checked-baggage weight. Bebop danced for passengers at the gate before boarding.

    Mozilla Meets Mythos
See All (40)

4.2
out of 5
143 Ratings

About

Twice weekly Technology News show covering the Tech Space. With Segments on Science, Mobile, Digital TV, FAA, FCC, Cyber Security, Gadgets and Tech Politics.

Information

  • Creator
    Todd Cochrane
  • Years Active
    2005 - 2026
  • Episodes
    40
  • Rating
    Clean
  • Copyright
    © 716264 Creative Commons Attribution-Noncommercial-Share Alike 3.0
  • Show Website
    Geek News Central Podcast

More From Tech Podcast Network

You Might Also Like