You're listening to Tech Beat, your daily look at what's moving in the world of technology.
OpenAI is facing serious questions tonight after researchers at Mindgard documented instances of ChatGPT spontaneously generating graphic violent and sexual imagery from a viral prompt. The findings point to a significant gap in the model's content filters — one that OpenAI has yet to fully explain. It's a reminder that guardrails in generative AI remain a work in progress, and the stakes for getting them wrong are real.
Closer to home for millions of Australians, Vodafone's mobile network went down today in what the company is calling intermittent service issues. No timeline for a full resolution has been confirmed, leaving customers in a frustrating holding pattern. Outages like this tend to sharpen the conversation around infrastructure resilience and just how dependent daily life has become on a handful of carriers.
And for anyone who has grown tired of Google nudging you toward its Gemini AI assistant every time you open a document, TechCrunch has published a walkthrough on how to disable those persistent prompts inside Google Docs. It's a small quality-of-life story, but it speaks to a broader tension — the line between a helpful feature and one that simply won't take no for an answer.
Keep surfing. Tech Beat out.
