It's standard practice to not let mutually-untrusting workloads share a CPU core. Spectre etc.

Which is why we should physically isolate these frontier training runs.

Run them in a data center in a faraday cage under a mountain.

The take home is that any shared resource is a potential communication channel between two willing participants. In this case it was CPU cache, presumably because openai didn't bother with basic vm security practices.

There's so many possibilities that this type of thing will never be solved by invisible (to the agent) boundaries. The sandbox is just another layer in w/e problem you've given it to solve.

Philosophical take home:

There is no such thing as isolation. We’re all connected. Even air gap conducts information in multiple ways, most famously used by humans evolved to use air vibrations to communicate. Photons and other subatomic particles already connects distant stars. We used those electrons and photons to create the modern telecommunications infrastructure. Just an additional layer of connectedness, but not the only one.