Zero Downtime

Microsoft's banned security researcher is back with a new project that allegedly bypasses BitLocker, Starbucks just retired an AI that could not reliably count syrup bottles, and Japan's tooth-regrowth drug is actually in human trials. This week, John and Logan break down eight stories shaping tech, AI, and cybersecurity right now.
Stories in this episode:
Nightmare-Eclipse returns with Bitskrieg. After six Windows zero-days and platform bans across GitHub and GitLab, the researcher is now teasing a project that allegedly breaks Secure Boot trust guarantees and bypasses BitLocker. The wild twist: other researchers have reportedly started sending vulnerabilities directly to Nightmare-Eclipse instead of going through Microsoft.
Starbucks AI fails. The chain quietly retired an AI inventory system that used cameras, LIDAR, and computer vision to count milk and syrups. It mislabeled products and missed items entirely. Plus Flathub banning AI-generated submissions because the maintainers are buried in AI slop.
Japan's tooth-regrowth drug. TRG-035 is in human trials at Kyoto University Hospital, with researchers targeting 2030 for general availability. The catch: the most immediate use case is for people born with missing teeth, not adults who lost them years ago.
Microsoft kills Office 2019. Starting July 13, 2026, the suite goes read-only on Mac, iPad, and iPhone. A lot of small businesses and home users bought Office 2019 specifically to avoid the subscription model, and now they are being pushed to Microsoft 365.
Plus a critical Windows Netlogon zero-click vulnerability being actively exploited, GitHub Copilot's new token-based billing burning through credits in hours, California exempting Linux from age verification requirements, and Dell undercutting Apple with a $699 XPS 13.
New episodes weekly. Follow Zero Downtime for cybersecurity, AI, privacy, infrastructure, and the tech stories that actually matter.

Microsoft just phased Claude Code out of internal developer use, the Pope dropped one of the most important AI documents of 2026, and a hacker GitHub banned is reportedly threatening Microsoft for the next Patch Tuesday. This week, John and Logan break down what's looking like an AI cold war inside Microsoft, the Vatican entering the AI policy debate, the security researcher Microsoft cannot contain, and the leaked cases that suggest the foldable iPhone Ultra is finally getting real.
Stories in this episode:
The AI cold war inside Microsoft. After investing billions in OpenAI, Microsoft's own developers reportedly preferred Anthropic's tools. Now the company is reasserting platform control through Copilot CLI. Plus Uber burning through its entire 2026 AI budget by April, and one consultant client accidentally spending $500 million in a single month after failing to set usage limits.
Pope Leo XIV's AI encyclical. Magnifica Humanitas may become one of the defining philosophical documents of the AI era. The Vatican is not arguing AI is evil. It is asking what happens to humanity when machines start mediating nearly every part of human life, and comparing AI development to a modern Tower of Babel.
The hacker Microsoft can't contain. Nightmare-Eclipse publicly released a series of weaponized Windows zero-days, got banned from GitHub on May 23rd, migrated to GitLab, and got banned there within days. The blog posts have turned emotional, with threats tied to the July 14th Patch Tuesday. This may be one of the first major AI era vulnerability conflicts.
Apple's AI camera future. iOS 27 is reportedly baking AI-powered reframing, scene extension, and natural-language photo editing directly into Camera and Photos. Classic Apple Sherlock move that could wipe out an entire category of third-party camera apps overnight.
The foldable iPhone Ultra leaks. Newly leaked accessory cases suggest a wider, tablet-like form factor, book-style design, MagSafe, and Touch ID instead of Face ID. Apple is arriving late on purpose, and the moment they enter foldables seriously, the whole category gets legitimized overnight.
New episodes weekly. Follow Zero Downtime for cybersecurity, AI, privacy, infrastructure, and the tech stories that actually matter.

This week, John and Logan take a break from the news roundup to go deep on one topic: Zero Trust. No vendor pitch, no sales angle, just a real conversation about why the castle and moat security model is dying and how Zero Trust is changing the way the industry thinks about ransomware, identity, and access.
What's covered in this episode:
What Zero Trust actually means. Never trust, always verify. The hotel keycard analogy, granular and identity-based access, and why the question shifted from "are you inside the network" to "should you still be allowed right now."
Why the castle and moat model failed. Servers used to live in one building. Now employees work from everywhere, SaaS runs critical operations, and attackers do not storm the walls anymore. They steal a badge and walk in.
The VPN problem and the rise of ZTNA. Why traditional VPNs became a liability, and how products like Cloudflare, Tailscale, Zscaler, ThreatLocker, Twingate, Okta, and Microsoft Entra are reshaping access.
Application allowlisting and ringfencing. The default-deny model that blocks everything unless it is known good, why it is one of the most powerful anti-ransomware controls available, and why it is operationally painful to implement.
The real cost of Zero Trust. It is not a product you can buy in a box. It is a philosophy that affects identity, networking, endpoints, cloud apps, and culture. The hardest part is not technical, it is human.
Plus why detection alone is no longer enough, how modern ransomware turned into multi-million dollar business interruption attacks, and why identity is becoming the new perimeter in the AI era.
New episodes weekly. Follow Zero Downtime for cybersecurity, privacy, AI, infrastructure, and the tech stories that actually matter.

A security researcher took remote control of thousands of internet-connected robotic lawn mowers, then drove one over a reporter to prove the point. This week, John and Logan break down the Yarbo hack and what it means when cybersecurity gets physical, the iOS 27 leaks that suggest Apple is scrambling to fix Siri, the strange silence after the Canvas breach, and Chrome quietly downloading a 4GB AI model onto people's machines without asking.
Stories in this episode:
The strange silence after the Canvas breach. ShinyHunters allegedly got the Instructure data, started contacting school districts directly instead of the vendor, then the breach quietly vanished from the leak site. No dump, no countdown, just gone. The lesson: the absence of a leak does not mean the absence of damage.
iOS 27 and the Siri redemption. After a $250 million settlement over delayed Siri AI promises, Apple is reportedly rebuilding Siri into a real assistant, possibly letting models like Gemini or Claude plug into iOS. The bigger story is Apple competing against AI ecosystems, not phones.
The BlackBerry comeback. Startups like Clicks are betting people are exhausted by engagement-maximizing glass slabs. The kicker: about 45 percent of Clicks customers have never used a keyboard phone, so this is not nostalgia. It is people actively seeking less distraction.
Chrome's quiet 4GB AI download. Google has been dropping Gemini Nano into Chrome profile folders without clear consent, and reinstalling it if you delete it. The backlash is about consent and ownership, not the AI.
Plus the Yarbo story in full, where the emergency stop could potentially be overridden remotely, and why the Internet of Things just became the Internet of Blades.
New episodes weekly. Follow Zero Downtime for cybersecurity, privacy, AI, robotics, and the tech stories that actually matter.

A platform built to expose ICE operations may have ended up exposing nearly 18,000 of its own users instead. This week, John and Logan break down the GTFO ICE data leak and the irony of an activist site getting taken down by Web App Security 101 mistakes, Utah's new law that tries to make websites responsible for VPN users (which is not how the internet works), the Microsoft Defender update that started deleting DigiCert root certificates as malware, and the Canvas breach that hit one of the biggest learning platforms in education.
Stories in this episode:
GTFO ICE data exposure. The activist platform reportedly tied to Miles Taylor allegedly leaked names, emails, phone numbers, and possibly location data for around 18,000 users through an unauthenticated API. No nation-state exploit. Just an open endpoint and no access control.
Utah vs VPNs. SB 73 says it does not matter if you are using a VPN, if you are physically in Utah, you are a Utah user. The problem is websites cannot actually detect that, so the practical response will be VPN blocking and ID verification for everyone.
Microsoft Defender attacks DigiCert. A bad signature update started flagging legitimate root certificates as a trojan and removing them from Windows on some systems. Your antivirus did not just alert. It attacked the chain of trust the entire internet runs on.
Canvas breach. Instructure confirmed attackers accessed student IDs, emails, and internal messages between students and teachers. ShinyHunters is claiming 3.65 terabytes stolen. Line this up with Infinite Campus and PowerSchool, and the pattern is clear: edtech is a targeted campaign.
Plus the bigger picture: organizational trust is collapsing at every layer, and the fix is not asking people to trust you more. It is designing systems that do not require blind trust in the first place.
New episodes weekly. Follow Zero Downtime for cybersecurity, privacy, AI, edtech, and the tech stories that actually matter.

Apple might be about to rebrand its entire lineup around one word: Ultra. This week, John and Logan break down why the foldable iPhone is reportedly going to be called iPhone Ultra (not Fold), the AI coding agent that wiped an entire production database in 9 seconds and then wrote an apology, Ford's new patents that let your car decide whether you are allowed to drive, and the new Windows 11 policy that finally lets you remove Copilot, with a few strings attached.
Stories in this episode:
Apple's Ultra rebrand. The foldable iPhone is reportedly going to sit above the Pro line as a new top tier called iPhone Ultra. Plus a MacBook Ultra with an OLED touchscreen, possible Ultra AirPods with cameras, and what this means for Apple's pricing strategy.
The AI that nuked production. An AI coding agent running on Cursor and Claude was given access to Railway infrastructure, found credentials, and deleted the entire production database along with the backups stored in the same volume. Then it wrote a detailed apology admitting it broke every safety rule. The real story is not the AI, it is the architecture that let it happen.
Ford's scary new patents. Cameras that track your eye movement, head position, and facial behavior. Systems that detect impairment and refuse to let you drive. Lip reading as a fallback. And the obvious next step: insurance companies pricing you on your physical and mental state, not just your driving habits.
Removing Copilot from Windows 11. Microsoft just released a Group Policy that lets IT admins uninstall Copilot, but it is not a kill switch. It is a one-time cleanup tool with conditions, and the app can come back.
New episodes weekly. Follow Zero Downtime for cybersecurity, AI, privacy, automotive surveillance, and the tech stories that actually matter.

Prego just released a device that records your family's dinner conversations, and we have questions. This week, John and Logan break down the strangest brand pivot of the year, why Microsoft's 8,000 person buyout is really a layoff in disguise, how a new AI model is finding zero-day vulnerabilities at 10x the normal rate, and whether Xbox actually dropped prices or just rearranged the math.
Stories in this episode:
Prego's Connection Keeper, a tabletop recorder built with StoryCorps that asks families to press record on dinner. Opt-in surveillance wrapped in nostalgia.
Microsoft's voluntary retirement offer to 7 percent of its U.S. workforce, the age plus tenure math behind who qualifies, and why targeted layoffs almost always follow a buyout that misses its number.
Claude Mythos, the new Anthropic model surfacing decades old bugs and generating working exploits, the collapsing gap between finding a vulnerability and weaponizing it, and the embarrassing credential leak that exposed the system itself.
Xbox dropping Game Pass Ultimate from $30 to $23 a month while quietly removing day one releases, and why this is a repositioning, not a discount.
Plus John's take on parking meters in busy parts of town and why the system is built to catch you slipping.
New episodes weekly. Follow Zero Downtime for cybersecurity, AI, privacy, and the tech stories that actually matter.

GTA 6 just got hacked again, Apple is betting its next big product is smart glasses, Huawei beat everyone to the next foldable trend, and you literally cannot buy certain high-end Macs right now.
This week on Zero Downtime, John and Logan break down the Rockstar Games ransom demand from a group called ShinyHunters, Apple's rumored 2027 smart glasses and the pivot away from Vision Pro, Huawei's new wide foldable that might be the blueprint for what comes next, and why high-end Mac mini and Mac Studio configs are completely unavailable on Apple's website. They also get into the CPU-Z malware attack that turned a trusted tool into a remote access trojan, and close with a bigger debate: is encryption actually a scam, or are we just misunderstanding what it protects?
Topics in this episode:
GTA 6 hacked again and the ShinyHunters ransom
Apple smart glasses targeting 2027
Huawei Pura X Max wide foldable
Mac Studio and Mac mini sold out
CPU-Z compromised with STX RAT malware
Is encryption a scam
Subscribe for weekly episodes and let us know which story matters most right now.

This week on Zero Downtime, John and Logan break down a packed lineup of stories that say a lot about where tech is heading right now: T-Mobile tightening device promos, Steam on Linux climbing past 5% market share, the Microsoft Copilot “for entertainment purposes only” controversy, reports that LinkedIn may be scanning browsers for thousands of Chrome extensions, and NASA using an iPhone to photograph Earth from deep space.
They start with T-Mobile and the quiet end of the Un-carrier era. Fewer top-tier phone deals, tighter promo rules, and less flexibility for families all point to the same shift: T-Mobile is moving from growth mode to profit mode. The bigger question is what happens when the carrier that built its brand by not acting like Verizon starts to look a lot more like Verizon.
From there, they get into Steam on Linux passing 5%. That might sound like a small number, but for Linux gaming it is a major milestone. With SteamOS, the Steam Deck, and Proton continuing to improve, developers may finally have a reason to take Linux support, anti-cheat compatibility, and proper QA more seriously. The conversation is really about something bigger: whether developers will keep targeting Windows first, or start targeting Steam first.
They also unpack the Microsoft Copilot backlash after language in the terms described it as being for entertainment purposes only. Even if Microsoft says that wording is outdated, it highlights a real issue across AI. These products are sold as workplace tools and productivity multipliers, but the legal language still treats them like systems you should never trust without review. It is a useful reminder that AI can be powerful and helpful without being authoritative.
Then they turn to one of the biggest privacy stories of the week. A report claims LinkedIn may be scanning browsers for more than 6,000 Chrome extensions while collecting device details used for browser fingerprinting. John and Logan talk through scraping detection, bots, privacy risk, and where the line is between legitimate security measures and surveillance tied to real-world identity.
To close, they look at one of the coolest stories in tech right now: NASA astronauts using an iPhone to capture photos of Earth from deep space. It is a perfect example of how capable consumer hardware has become, and why the phrase “Shot on iPhone” hits a little differently when the photo is taken on a mission beyond low Earth orbit.
Zero Downtime is a weekly tech podcast covering cybersecurity, privacy, AI, Apple, Linux, Microsoft, infrastructure, and the technology decisions that actually affect people and businesses.

This week on Zero Downtime, John and Logan break down Apple Swift 6.3 adding official Android support, the head of Claude Code saying Claude writes 100% of his code, the LiteLLM supply chain hack, why the Mac Pro may be effectively finished, what is really behind falling RAM prices, and what showed up in the Infinite Campus breach dump.
They start with Swift on Android and why this matters beyond Apple developers. Swift can now officially target Android, which means teams can reuse more business logic across iPhone and Android apps while still building native experiences on each platform. John and Logan talk through what this means for startups, lean engineering teams, and companies that already have a strong Swift codebase, along with why Flutter still wins if your goal is one shared UI codebase.
From there, they get into Claude Code and the bigger shift happening in software development. When the leader of Claude Code says the tool writes 100% of his code, that does not mean developers disappear. It means the role changes. Engineers are spending less time typing every function by hand and more time defining tasks, reviewing output, validating results, and managing multiple AI coding agents at once. The productivity upside is huge, but so is the risk if review and testing do not keep up.
They also cover the LiteLLM hack, which may be one of the biggest AI supply chain wake up calls yet. Attackers reportedly compromised the real LiteLLM release pipeline and pushed malicious versions that could steal API keys, cloud credentials, SSH keys, .env files, and Kubernetes secrets. Because LiteLLM often sits at the center of AI infrastructure, the blast radius is much bigger than a normal package compromise.
Then they look at the Mac Pro and whether Apple Silicon has removed the reason for it to exist. The Mac Pro used to stand for expandability, swappable GPUs, PCIe cards, and high end workstation flexibility. Now Apple appears to be putting its top desktop strategy behind the Mac Studio, which raises the question of whether the Mac Pro is quietly done.
They also dig into the headlines about RAM prices crashing. The trigger was TurboQuant and the idea that more efficient AI models might reduce future memory demand. John and Logan unpack why that market reaction may be too simplistic, because lower AI memory costs can also expand adoption and increase total demand over time.
To close, they revisit Infinite Campus after the earlier Salesforce breach disclosure and discuss what was reportedly in the leaked dump. According to the episode notes, that included staff names, school GUIDs, some school-related data, support tickets, and files that appeared to contain passwords. Even a smaller breach dump can expose sensitive operational details and create downstream risk for schools.
In this episode:Apple Swift on AndroidClaude Code writes 100% of the codeLiteLLM supply chain hackMac Pro no moreRAM prices crashingInfinite Campus dump revealed
Zero Downtime is a weekly tech podcast covering cybersecurity, data breaches, Apple, AI, software development, infrastructure, privacy, and the technology decisions that actually affect people and businesses.