There is a widening readiness gap in enterprise security as the rapid adoption of generative AI (genAI) outpaces defenders’ ability to secure it, a new report found.

Along those lines., 36 percent of security leaders and practitioners said genAI is moving faster than their teams can manage, a sobering reality as organizations continue to embed AI deep into core business operations, according to penetration testing as a service (PTaaS) provider, Cobalt’s State of LLM Security Report 2025.

Meanwhile, with this concern, 48 percent of respondents believe there needs to be a “strategic pause” to recalibrate defenses against genAI-driven threats.

“Threat actors aren’t waiting around, and neither can security teams,” said Gunter Ollmann, CTO at Cobalt., “Our research shows that while genAI is reshaping how we work, it’s also rewriting the rules of risk. The foundations of security must evolve in parallel, or we risk building tomorrow’s innovation on today’s outdated safeguards.”

Report Highlights

Other key findings from the report include:

Schneider Bold
  • 72 percent of respondents cite genAI-related attacks as their top IT risk, but 33 percent are still not conducting regular security assessments, including penetration testing, for their large language model (LLM) deployments.
  • 50 percent of respondents want more transparency from software suppliers about how they detect and prevent vulnerabilities, signaling a growing trust gap in the AI supply chain.
  • Security leaders (C-suite and VP level) remain concerned about long-term genAI threats like adversarial attacks (76 percent) versus the 68 percent of practitioners which expressed the same concern. However when it came to near-term operational risks such as inaccurate outputs, 45 percent of practitioners expressed concern versus 36 percent of security leaders.
  • Top concerns among all survey respondents include sensitive information disclosure (46 percent), model poisoning or theft (42 percent), and training data leakage (37 percent), all pointing to an urgent need to protect the integrity of data pipelines.
  • Overall, 69 percent of serious findings across all pentest categories end up resolved but this falls to just 21 percent of the high-severity vulnerabilities found in LLM pentests. This is a concern given 32 percent of LLM pentest findings are serious and is the lowest resolution rate across all test types conducted by Cobalt.

“Much like the rush to cloud adoption, genAI has exposed a fundamental gap between innovation and security readiness,” Ollmann said. Additionally, “mature controls were not built for a world of LLMs. Security teams must shift from reactive audits to programmatic, proactive AI testing – and fast.”

Click here to register to download the full report.

Pin It on Pinterest

Share This