FRONTIER · GLOBAL
Microsoft publishes AI code of conduct limiting model autonomy in financial and security contexts
Microsoft released a 'Humanist AI' framework instructing its models to support humans rather than replace them, with specific safety constraints on hacking, deception, and autonomous financial decisions. Reflects broader industry push to throttle frontier model capabilities.
WHY IT MATTERS
Sets precedent for self-regulation by major AI labs; signals to regulators and enterprises that guardrails on model behavior are operationalizable, not theoretical.
Source: CNBC · 2026-09-14