How Agentic AI Deployments Open Doors to Classic Cyber Attacks

The article discusses the risks associated with deploying agentic AI systems, emphasizing that the technology itself isn't inherently risky, but rather the way organizations implement it can lead to vulnerabilities. Research shows that many developers are hastily adopting AI without understanding the underlying technology, leading to traditional vulnerabilities in their coding.

Experts from Acronis highlight that agentic AI systems are not black boxes; they consist of both deterministic software and non-deterministic AI components, and vulnerabilities often arise at this intersection. Examples of critical vulnerabilities in AI-enabled platforms like Salesforce and ServiceNow illustrate these risks. To secure agentic technology, the article advocates applying traditional cybersecurity principles alongside understanding the AI's unique features.