OpenAI's Model Cracked a Math Problem That Stumped Humans for 80 Years

OpenAI's Model Cracked a Math Problem That Stumped Humans for 80 Years

A mathematician's whiteboard doesn't usually look like a crime scene, but that's the mood right now: according to Ars Technica, an OpenAI model has solved a famous problem that resisted human proof for eight decades. The details matter. This isn't "the model wrote a clever paragraph." It produced a result researchers could check step by step — which is what separates a flashy demo from a genuine scientific event.

That distinction is everything. AI has been embarrassingly good at sounding right for a while. Being right in a field where every step must survive hostile peer review is a different sport entirely. If this holds up, it suggests these systems are moving from "assistant that drafts and summarizes" toward "tool that advances the frontier" — especially in math-heavy science, where the bottleneck is often not raw computation but the ability to navigate a mountain of possible approaches.

A verified solution to a famous problem doesn't just earn bragging rights. It gives researchers a new question: what else has the model cracked that humans stopped attempting years ago? The weird part is no longer that the machine can talk like a mathematician. It's that it may have done something mathematicians failed to do for 80 years. That's not a headline. That's a warning label.

Gobble's Take: AI is no longer just summarizing the world — it's starting to rewrite the parts humans gave up on.

Source: Ars Technica

Strava Is Locking Down Its Data Before Wall Street Decides It's a Gold Mine

A runner uploads a workout, and somewhere a bot is copying that data a million times over. That's the fight Strava is now picking: according to TechCrunch, the fitness app is cracking down on scrapers — automated tools that vacuum up public or semi-public data at scale — ahead of its expected IPO.

This matters because Strava isn't just a running app anymore. It's a living map of how millions of people move, train, and commute. That data is valuable on its own, and worth dramatically more once Wall Street starts pricing the company for public markets. If you're a user, this is the tell: a free-ish product starts acting like a company preparing for scrutiny — tighter rules, more lockdowns, less "sure, take whatever you want."

The subtext is one every consumer internet company eventually learns the hard way: if your users create the treasure, someone will build a machine to steal it. Strava is saying no before investors ask why it waited. The workout data is now an asset class, and the party for bots is over.

Gobble's Take: The closer an app gets to an IPO, the more it starts treating your habits like inventory it owns.

Source: TechCrunch

Google Is Opening Its First Store Outside the U.S. in Tokyo This Summer

A shopper in Tokyo will soon be able to walk into a Google store — the first time Google has put a physical flagship anywhere outside the United States. Engadget and 9to5Google report the store opens this summer, giving Google a stage where its hardware, software, and brand all share the same room for the first time in an international market.

For Google, this is bigger than display tables and demo units. It's a direct play at the premium-device theater Apple has owned in Japan for years. Google wants customers to hold a Pixel, try the earbuds, and feel the ecosystem before they check a spec sheet — because that five-minute in-store experience closes sales that no ad ever does. The Tokyo location also puts Google's AI features front and center in one of the world's most tech-forward consumer markets.

The timing is deliberate. Google is pushing harder on hardware and premium devices at the exact moment it needs to look like a serious lifestyle brand, not just a search engine with a phone on the side. The Tokyo store won't just move units — it will sell the idea that Google belongs in your pocket, your home, and your Saturday afternoon. Silicon Valley has figured out that sometimes the best interface is a front door.

Gobble's Take: Google opening a physical store in Tokyo isn't retail strategy — it's a declaration that the hardware wars are being fought in person now.

Sources: Engadget · 9to5Google

China Just Approved the World's First Invasive Brain-Computer Chip

A patient lying still in a clinic does not sound like the start of a consumer-tech story, but here we are. MIT Technology Review reports China has approved the world's first invasive brain-computer interface chip. "Invasive" is the word that matters: this is not a headset you strap on. It means hardware placed into or directly onto the body to read neural signals — which is precisely why the jump from laboratory prototype to regulatory approval is such a significant threshold.

The promise is real and unsettling in equal measure. If it works as intended, the technology could help people who cannot move or speak interact with computers in ways that currently exist only in science fiction. But once a government starts approving implants this close to human cognition, the conversation shifts quickly from engineering to ethics: who owns the data when the data originates in your nervous system, and who decides what the chip records?

The rest of the world will watch closely — not because everyone wants hardware in their skull, but because approvals create momentum. First comes medical use. Then comes the argument for commercial use. Then comes the moment when your employer, your insurer, or your phone manufacturer starts asking whether you'd like to interface directly. That's not a distant hypothetical. That's a policy draft with a pulse.

Gobble's Take: The moment a brain chip gets approved anywhere, "privacy" stops meaning your passwords and starts meaning your thoughts.

Source: MIT Technology Review

Nvidia's RTX Spark Wants to Turn Your Laptop Into a Personal AI Supercomputer

A laptop sitting on a desk in mid-2026 is no longer just a laptop. Nvidia has announced RTX Spark — a chip it describes as "the most efficient PC chip ever built," according to The Verge — aimed at bringing serious AI processing to Windows laptops and desktops without the noise, heat, and battery drain that have made AI workloads a desktop-only affair.

The underlying bet is that the next wave of AI won't only live in cloud servers owned by a handful of giant companies. It will move to the edge: the machine in your backpack, on your kitchen counter, under your arm on a commute. For developers, creators, and anyone who wants fast local AI without routing every file through someone else's server, that's a meaningful shift. The first RTX Spark laptops are already confirmed by The Verge, meaning this isn't a roadmap slide — it's a product arriving on shelves.

Nvidia is no longer just selling to server farms and gaming rigs. It's positioning the personal computer itself as an AI appliance, which is a compact way of saying your next laptop will be judged less by how fast it opens a browser and more by how well it runs models, edits video with AI, and generates content locally. The humble notebook just got conscripted into the compute arms race — whether you asked for it or not.

Gobble's Take: If Nvidia pulls this off, the gap between "my laptop" and "AI machine" won't just blur — it'll disappear on your next upgrade cycle.

Sources: The Verge · Engadget

Quick Hits

• South Korean rocket startup Unastella raises $24M after launching from a garage: The company, which reportedly conducted early rocket tests from a residential property, closed a $24 million round as it scales toward orbital launches. TechCrunch
• OpenAI Codex authentication tokens stolen in npm supply chain attack: A malicious package called codexui-android was planted in npm, the JavaScript package registry, to harvest developer credentials tied to OpenAI's coding tool. The Hacker News

In Case You Missed It

Yesterday's top stories:

• Connecticut’s AI rulebook is getting real, fast
• Anthropic and OpenAI: product-market fit, or a trillion-dollar bill
• Pope Leo XIV’s AI encyclical is not subtle about the human cost