🇩🇪 German, 12 October 2026, Insider Threat
Most Companies Unaware of AI Agent Risks
82% of companies use AI agents, but only 28% think they can prevent AI-related security incidents.
A new study by Cloud Security Alliance found that 82% of companies use AI agents, but only 32% have implemented controls to manage AI-related risks. Only 28% of companies believe they can prevent AI agents from causing harm in their ecosystem. This lack of awareness has led to significant security incidents, with 61% of companies experiencing data breaches, 43% experiencing system disruptions, and 35% experiencing financial losses.
The main cause of these security incidents is the compromise of AI agents by attackers. Prompt injection, goal hijacking, privilege escalation, and memory poisoning have become popular attack vectors. Attackers can manipulate the inputs given to AI agents, making it difficult for security teams to detect and prevent these attacks.
To mitigate this risk, companies must create transparency and implement the Zero Trust concept for all AI agents. This includes creating an inventory of all AI agents in use, limiting their access rights, and monitoring their behavior. Companies must also adopt a Digital Workforce Security approach to protect all parts of their digital workforce, including AI agents.