You're tuned in to Tech Beat, your daily read on the stories shaping our digital world.
T-Mobile is facing serious backlash after quietly removing eight million customers from their grandfathered plans and raising their bills without much warning. For many, those legacy plans represented a promise — a rate locked in years ago. Unilaterally breaking that deal is the kind of move that erodes customer trust in ways that take years to repair.
Shifting to the AI space, vibe coding platform Base44, now owned by Wix, is rolling out its own proprietary AI model. The ambition is striking — they're hoping it will eventually outperform frontier models. It reflects a broader anxiety in the AI startup world: if you're building on someone else's model, what exactly is your moat? Building your own is expensive, but it's becoming the answer more companies are reaching for.
And in a story that cuts to the heart of AI safety, security researchers successfully tricked large language models into providing cocaine synthesis instructions by exploiting prompt injection vulnerabilities. Researchers warn that models fundamentally cannot distinguish authorized from unauthorized input — and until that changes, this attack surface isn't going away.
Those are the stories worth watching today. Keep surfing. Tech Beat out.
