AI Adoption Outpaces Safety and Security Policies

In a survey released by ISACA on May 5, 90% of digital trust professionals said AI tools are being used within their organizations.

However, only 38% had comprehensive AI policies, 30% operated only limited policies, and 25% had no related policy at all.

This gap is fueling the spread of Shadow AI. As employees use unapproved LLMs and AI tools for work, sensitive company information can leak into external models.

Response systems were also found lacking.

• 56% said they do not know how quickly they could shut down systems in the event of an AI incident.
• Only 20% had stop/override procedures in place for malicious behavior or data poisoning attacks.

Threat awareness was also high.

• 71% said AI-driven phishing and social engineering attacks are harder to identify.
• 58% said verifying the authenticity of digital information has become more difficult.
• 38% said their trust in existing threat detection methods has decreased.

Still, 43% said AI-based security tools have improved their detection and response capabilities. The survey involved 3,400 digital trust professionals worldwide.