AI safety is falling behind the pace of development

Businesses need continuous monitoring, stronger guardrails and human oversight to keep increasingly capable AI systems under control.