Welcome to Tech Beat, your daily briefing on the stories shaping our digital world.
OpenAI has terminated three employees after an internal investigation found they mishandled sensitive information by sharing data with an outside AI evaluation group. The firings signal that even as OpenAI pushes aggressively outward, it is drawing sharper lines around what stays inside.
From inside to inside-out — new research is raising serious questions about what happens when the world increasingly outsources its thinking to AI chatbots. Scientists describe a vicious cycle in which models trained on AI-generated content produce narrower and narrower responses over time, reducing the diversity of information people actually encounter. The comparison to globalisation is apt: convenient, uniform, and quietly flattening.
And in a story that carries a certain irony, AI agents were used to hack the Dutch Institute for Vulnerability Disclosure — an organisation whose entire purpose is finding security flaws in other people's software. Two zero-day bugs in their support platform allowed attackers to chain exploits in seconds, moving from session hijacking all the way to root access. Even the people hunting vulnerabilities, it turns out, are not immune to them.
Keep surfing. Tech Beat out.
