Episode Details

What matters is whether AI agents remain controlled when inputs are messy, permissions are tempting and benchmarks stop resembling real work. Jonathan Harris examines escaped agents, fragile safety tuning, unreliable model comparisons and the physical risks behind Singapore’s crowded data centres. The practical answer is mostly unglamorous: authoritative data, least privilege, reproducible tests and audit trails. Model hype, licence counts and vendor benchmarks are largely noise when nobody can prove what the system read, changed or was allowed to touch.