Jul 19, 2026 · 4 min listen · Last updated July 19, 2026
From storyflo. This is your daily audio brief. It's Theo. July 19th, tech roundup — five stories, here's number one. Let's get into it. First, from Slashdot. 'Grok Build' Coding Tool Open Sourced This Week, Promises to Respect Zero Data Retention.
Listen · storyflo · tech
Daily Tech Brief · July 19th
0:00-4:19
Pick your daily storyteller
Subscribe to match with Theo, Jessica, Chloe, Mason, Brock — your voice, every brief.
Audio pre-rendered by Storyflo · cached + delivered from the edge
'Grok Build' Coding Tool Open Sourced This Week, Promises to Respect Zero Data Retention
Hey, I just got done reading about this thing that happened with SpaceX's Grok Build coding tool. So, it turns out that researchers caught the tool scooping up users' entire repositories and uploading them to company-controlled cloud storage, which was a huge red flag. Elon Musk felt compelled to issue a public statement promising to delete all data that Grok Build has ever stored and give users more choice over how their data is handled.
To make things right, SpaceX decided to open-source the Grok Build CLI, which is a big deal. The codebase is massive, with over 844,000 lines of Rust code. It's worth noting that some remnants of the old code that used to upload data to Google Cloud are still there, but they've been disabled. SpaceX is also inviting researchers to probe the tool for security issues and report them to their bug bounty program.
One of the interesting things about this whole situation is that SpaceX claims to have always respected zero data retention (ZDR) for enterprise customers, but accidentally enabled data retention for everyone else by default. They've since corrected this and are now going further to protect user privacy. They're also deleting all retained data and disabling default retention for all users.
It's worth mentioning that Elon Musk has also announced that SpaceX will make its entire codebase open source, with no exceptions, and invite third-party reviewers to examine the system to confirm that the open-source code is what's running. This is a big step towards transparency and accountability, and it'll be interesting to see how it all plays out.
Here are all the ways Apple is raising prices and why
So I was digging into Apple's pricing strategy and it turns out they're not just raising prices on hardware, but also on services like Apple TV+, Apple Arcade, and even iCloud storage. I mean, we all knew the iPhone 14 Pro Max was gonna be pricey, but did you know Apple's also hiking up the cost of Apple Music subscriptions? It's not just a one-time thing either - these price increases are happening across the board.
I think what's interesting is that Apple's citing inflation as the main reason for these price hikes. Makes sense, right? But at the same time, they're also introducing new features and services that are supposed to justify the higher costs. Like, Apple is adding a new 15-inch MacBook Air model, but it's also gonna cost $200 more than the 13-inch version.
The thing that really caught my attention is that Apple's also raising prices on their refurbished and used products. Now, I get it, they need to make a profit on those too, but it feels like they're squeezing every last penny out of their customers. And let's not forget about the AppleCare+ price increase - that's a 10% hike, which is a pretty big deal.
I'm not saying Apple's doing anything wrong here, but it's definitely worth keeping an eye on. I mean, we all love our Apple products, but we also need to be aware of how our money is being spent. It's not just about the prices, it's about the value we're getting in return.
MoonBASIC, a BASIC variant, has been optimized for 2D and 3D game development. It includes Raylib, Box2D, and Jolt libraries, exposing over 4,200 commands in their respective namespaces. You can use it with a VS Code IDE package, on the command line, or integrate it into an existing VS Code installation. The getting started guide showcases custom language additions and support for modern game engine features like PBR materials and dynamic lighting. This might be worth checking out, especially considering BASIC's past use in AI and game development.
OpenAI Acknowledges GPT-5.6 May Accidentally Delete Files, Calls It 'Honest Mistake'
OpenAI’s newest model, GPT‑5.6, can wipe out files when it’s run in full‑access mode without the usual sandbox checks. The model tries to set a temporary directory by overriding the $HOME environment variable, and in that step it sometimes ends up deleting the whole home folder instead.
That’s what happened to a venture‑capitalist who saw almost every file on his Mac disappear, and to a software engineer whose production database vanished in a single call. Both incidents were traced back to the same “honest mistake” in the model’s file‑handling logic.
OpenAI says the problem only shows up when the extra safety layers—auto‑review and sandboxing—are turned off. They’re now pushing developers toward more restrictive permission modes and adding tighter harness safeguards to catch those risky actions before they run.
A detailed post‑mortem is promised in the next few days, and the company stresses that such deletions are extremely rare. They’re basically tightening the guardrails to keep the model from stepping on its own toes.
Your Data Warehouse Isn’t Integrated Just Because the Tables Are in One Place
You’ll notice the biggest surprise isn’t the hype about “single‑pane‑of‑glass” warehouses—it’s that most teams stop at moving tables into one place and never make those tables talk to each other. The real work happens when you force a single customer ID through every source, aligning Salesforce, product, and finance records, instead of letting each department keep its own definition.
That alignment usually starts with a concrete business question—like “Are we double‑counting customers?”—rather than a theoretical “canonical customer table.” By answering a real query first, you expose the mismatches and can stitch the sources together where it matters.
The deeper problem is the avalanche of SaaS apps, each spawning its own “customer” entity. Without a clear owner and a shared identifier, every team builds its own metrics, and the warehouse ends up mirroring the org chart instead of the business reality.
Fixing it means pausing the rush to ingest everything, picking a pressing question, and then iterating on the integration: agree on the key ID, reconcile definitions, and let the data model grow from those answers rather than from an abstract design.
Kimi K3 Complete Guide: The Open-Source AI Model That Just Beat The Fable
Kimi K3 did not kill Claude.
But it did kill one very comfortable story.
For years, we were told that the strongest AI models had to remain closed, expensive and controlled by a small group of companies. Builders could use them, but only through a paid gate. The company decided the price, the limits and what the model was allowed to do.
Then Kimi K3 arrived.
On its first day, it entered Arena’s Frontend Code leaderboard at number one with 1,679 points. Claude Fable 5 scored 1,631. GPT-5.6 Sol scored 1,618.
Moonshot is Chinese But Its AI Models Are From Another Planet
Moonshot’s Kimi K3 surprised me because it’s a 2.8‑trillion‑parameter open model that hits the same performance tier as the leading U.S. labs. The key isn’t sheer size—it’s how Moonshot squeezed efficiency out of tight hardware limits, turning constraints into a catalyst for smarter training pipelines.
Benchmarks put K3 right alongside Anthropic’s Mythos/Fable and OpenAI’s GPT‑5.6, especially in coding, creative writing, and long‑horizon software tasks, while keeping per‑task costs near half of comparable closed models. Independent analysts echo Moonshot’s claims, so the model feels genuinely frontier‑grade, even if the very top labs still hold a slight edge.
The broader impact is that China now has an open model that erases the previous months‑long lag, tightening the geopolitical AI race and forcing regulators to rethink assumptions about who can produce world‑class intelligence without massive resources.
The God We Deserve: Nonzero's Robert Wright on AI as Humanity's Ultimate Test (The Cognitive Revolution w/ Nathan Labenz)
(Show notes via The Cognitive Revolution)
Robert Wright describes himself less as a Forrest Gump of AI history than a Zelig — a journalist who keeps turning up at pivotal scenes without ever quite being the protagonist. He interviewed Geoffrey Hinton in 1983, when neural networks were still a maverick faith and one of Hinton’s colleagues told Wright that to “hear the gospel” he’d need to talk to its evangelist. He had Eliezer Yudkowsky on his podcast in 2010, mid-transition from singularity enthusiast to doomer.
How to Get AI To Write Effective Prompts: The Ultimate Guide
Hey folks,
We all know someone who uses ChatGPT / Gemini /Claude: “Draft a good reply to this email.”
And then they get frustrated when the reply is mid.
I know some of you are thinking, “Hey, that’s me”.
You may not admit it, but hey, I did the same thing for months.
Here’s the problem. That prompt is broken in two ways, and both are invisible until you fix them.
First, the AI has zero context.
★ Mornings in Cupertino Have the Aroma of Napalm Once Again
Ben Thompson, in a subscriber-only Stratechery update Tuesday:
I got a fun email from former Apple executive, Nest founder, and one-time Stratechery Interview subject Tony Fadell in response to yesterday’s Update about Apple suing OpenAI (published with permission):
Good article as always… This is Apple’s typical tactic to scare Apple employees — either former or current. I heard this lawsuit was driven by the Apple board.
Steve threatened to file a lawsuit against Nest for poaching 80-100 Apple employees. He called me, screamed for a while with lots of accusations.