Shadow AI Crisis: Businesses Lose Track, Costs Soar

A new report shows a growing problem with "shadow AI" in businesses. This means organizations are losing track of their AI software components. What's interesting is that in 2025, one in five organizations reported a security incident linked to shadow AI. Those with high levels of shadow AI paid an average of $670,000 more per breach. The report also finds that 61% of new enterprise applications are designed with AI components. However, 62% of security professionals admit they can't tell where large language models are deployed in their organizations. Organizations are already facing real security incidents. For example, 76% have experienced prompt injection incidents, and 66% have been hit by vulnerable LLM code. These are not future risks, but active problems right now. The bottom line is that the speed of AI innovation is creating a security visibility gap, making it harder for companies to defend themselves.