Video

Unknown · 0:00

The talk argues that safe agent autonomy with free choice of models and harnesses requires governing a common runtime (a sandbox with deterministic, fine-grained, semantic policy) rather than trusting the model, the h...

Read the full summary on tuber