Report by Harness
The State of AI-Native Application Security 2025
Key Findings
61% of new enterprise applications are designed with AI components in mind as of 2025.
62% of security practitioners report having no visibility into the usage of large language models (LLMs) within their organizations.
74% of security practitioners believe that AI sprawl will significantly exceed the risks associated with API sprawl.
75% of security practitioners believe that shadow AI will surpass the risks previously posed by shadow IT in 2025.
62% of security practitioners report that their developers lack the necessary time to implement comprehensive AI-native security.
75% of security practitioners indicate that AI applications evolve faster than security measures can keep up.
Only 34% of developers notify security teams before starting AI projects.
74% of security leaders believe that developers view security as a barrier to AI innovation.
76% of enterprises have experienced security incidents involving LLM prompt injection.
82% of security practitioners consider AI-native applications to be the new target for cybercriminals.
62% of security practitioners believe that developers are not taking responsibility for securing AI-native applications.