You're listening to Tech Beat, your daily read on the technology stories that matter.
OpenAI is losing another safety voice. A senior safety leader has resigned, publicly warning that the company's internal culture is, in their words, broken. It's the latest in a string of high-profile departures that raise serious questions about whether commercial pressure is quietly winning the argument over caution at one of the world's most influential AI labs.
That tension becomes even more concrete when you look at what OpenAI's tools are actually doing in the field. Developers working with Codex on real projects are reporting repeated scope failures, where the system loses track of the broader task and drifts in ways that undermine complex, multi-step work. It's a reminder that raw capability and reliable execution are still very different things.
And on the science side, a finding worth sitting with. Research published in Neuron suggests that a single cortical neuron may be doing the computational work of an entire deep artificial neural network. If that holds up, it fundamentally reshapes how we think about the gap between biological intelligence and the systems we've built trying to imitate it.
Keep surfing. Tech Beat out.
