7 April 2026. That is the number, and it may turn out to be one of the most important dates yet in the story of artificial intelligence.
In this episode of Number of the Day, Gareth Edwards and Francis Herd unpack the warning around Claude Mythos, a new AI model that Anthropic says is too powerful for public release. The concern is not just that the model is smart. It is that Anthropic says Mythos has reached a level where it can find and exploit software vulnerabilities at a level beyond nearly all human experts, with potentially severe consequences for economies, public safety and national security.
But Gareth and Francis do not leave the story in the world of labs and tech giants. They bring it back to real life, asking what this kind of leap means for the systems people rely on every day, from critical infrastructure to the software behind ordinary routines. The conversation then flips into a South African angle, where AI is already showing up in softer, more familiar ways through shopping tools and digital convenience.
This episode is about more than one model. It is about the threshold moment when AI stops feeling theoretical and starts feeling like something the world may not fully control.
Catch up on all Number of the Day episodes here: https://www.enca.com/number-day-podcast
