This Week in Tech - Leaks, Subscriptions, and the Rules Catching Up

This week didn’t scream for attention. It didn’t need to. Policy tweaks. Product changes. Old problems showing up in bigger numbers. Here’s what actually mattered, once you zoom out a little.

149 Million Login Credentials Spill Out Again

A dataset containing roughly 149 million usernames and passwords surfaced online, stitched together from older breaches and newer leaks. No single company caused it, which is almost the point. This is what years of recycled passwords and breached services look like when they pile up. We’ve been here before, many times, and the numbers keep getting bigger. The takeaway hasn’t changed much since the early 2010s: reused credentials age badly, and attackers are patient.

Tesla Pushes Autopilot Further Behind a Paywall

Tesla moved more Autopilot features into a monthly subscription, continuing a shift that started years ago when cars became software platforms. Hardware stays in the vehicle, capability lives behind a login. This isn’t new for Tesla or the auto industry anymore, but it’s still a reminder that buying the car isn’t the same as owning everything it can do. Transportation is slowly inheriting the economics of apps.

Threads Quietly Crosses 400 Million Users

Meta’s Threads passed 400 million monthly active users without much noise. No big redesign, no cultural reset. Just steady growth tied closely to Instagram’s gravity. It’s a familiar Meta move: ship, connect it to an existing network, and let scale do the work. Whether Threads becomes a true public square or just a calmer side room is still an open question.

TikTok’s US Future Gets Another Structural Rewrite

ByteDance agreed to a new proposed US structure for TikTok as regulators continue circling. This is the latest chapter in a story that’s been running since 2020, where national security concerns collide with global platforms. The details keep changing, but the pattern doesn’t. Governments want leverage. Platforms want access. Users mostly just want their feeds to load.

How People Are Actually Using AI at Work

New data on real AI usage showed something refreshingly boring. People are using AI to summarize, draft, brainstorm, and clean things up. Not to replace themselves, not to build sci-fi futures overnight. Just to move faster through familiar tasks. It’s less about transformation and more about friction removal, which is usually how technology actually sticks.

AI Regulation Starts Looking Real Worldwide

Countries rolled out clearer AI rules for 2026, and for once they weren’t all wildly different. Transparency requirements, risk categories, and data protections are starting to rhyme across regions. This mirrors what happened with privacy law a decade ago. First confusion, then alignment, then compliance tools everywhere.

VS Code Shows Up in Active Cyberattacks

Security researchers documented attacks abusing Visual Studio Code features to move quietly inside systems. The editor itself isn’t compromised. It’s trusted software being repurposed. This kind of thing has happened before with PowerShell and admin tools. Attackers follow trust. Developers just happen to live there.

OpenAI, Microsoft, and Musk Keep Lawyers Busy

Elon Musk’s lawsuit against OpenAI and Microsoft pushed old disagreements back into public view. The dispute isn’t really about one product or one model. It’s about governance, control, and how nonprofit origins collide with commercial scale. AI history is being written in court filings now, not just research papers.

Looking Ahead

Nothing here felt flashy, and that’s the signal. Security problems compounded. Software subscriptions spread. AI rules hardened. Platforms adjusted quietly. This is the part of tech history where the foundations settle. The effects tend to show up later.