- Microsoft chief executive Satya Nadella has stated that all artificial intelligence models should be treated as compromised and potentially dangerous.
- Nadella advocated for built-in emergency brakes allowing authorised individuals to pause or terminate AI models mid-task.
- Microsoft president Brad Smith called for urgent global safety rules, noting that AI is currently less regulated than everyday consumer products such as milk and washing machines.
- Both executives emphasised the necessity of an engineering approach to safety, including standardised containment technologies for advanced systems.
- The proposals align with broader tech industry calls from companies like Google DeepMind and Anthropic to develop AI kill switches following recent security incidents.
IN FULL
