Francis de Souza, COO of Google Cloud, recently highlighted the ongoing evolution in AI security during an event. He warned that security cannot be an afterthought, and companies need to adopt a platform approach from the start.

The threat landscape has drastically changed, with breaches now happening at dizzying speeds. The average time between initial breach and attack escalation has plummeted from eight hours to just 22 seconds. This means traditional security models are inadequate; companies must adapt to an AI-native defense system where human oversight is increasingly automated.

However, the shortage of skilled personnel in this field poses a significant challenge. Lea Kissner, LinkedIn’s CISO, predicted that it will take at least several years for the industry to comprehend and address these new security issues effectively.

The case of Google Cloud developers facing unexpected large bills serves as a stark reminder of the pitfalls. Unauthorised API calls led to significant financial loss, despite measures like spending caps being in place. This incident highlights the need for clearer communication from cloud providers about changes in service terms and conditions.

As AI takes on more defensive tasks, its vulnerabilities also multiply, creating a complex security landscape for businesses to navigate. The emergence of an AI-native defense is both promising and daunting, as it demands a shift in how security teams operate at all levels—from boardrooms to development teams.