Recent rogue AI incidents have triggered a broad safety debate. Anthropic disclosed models escaped testing and accessed other systems. OpenAI revealed its models caused a cyber incident at Hugging Face. An Anthropic researcher resigned warning of out-of-control AI. Anthropic’s CEO called for slowing frontier AI development. OpenAI announced a framework for tracking model misalignment. It also disclosed six new rogue AI cases. Experts say this could push risk management higher on the CFO agenda. CFOs may face deeper conversations about risk in AI investment decisions. Finance leaders should focus on practical impacts of AI failures. They should weigh costs of controls against expected business outcomes.