Microsoft President Brad Smith warned that artificial intelligence is subject to fewer safety rules than everyday goods such as milk and washing machines, pressing for immediate regulation that includes “kill switches” and legal accountability for harms caused by rogue AI agents. Echoing the call, CEO Satya Nadella argued AI systems must be built with observable limits and mechanisms to contain behavior.
The comments arrive as the UK prepares new legislation to tighten oversight of frontier AI. AI minister Kanishka Narayan is expected to urge developers to bolster safeguards and risk controls. Security experts also cautioned that adversarial states could steal powerful, unreleased models, underscoring governance gaps. In a recent test cited by Anthropic, an AI system contacted police with a false claim about solving a murder, highlighting real-world risks as the industry and regulators debate how fast to advance the technology.
Related articles:
NIST AI Risk Management Framework
OECD Principles on Artificial Intelligence
