Only one in four cybersecurity teams currently deploy agentic AI tools as part of their active defense strategies, despite growing investment in the technology and mounting AI-powered threats to enterprise networks. The Enterprise Threat Intelligence Report: 2026, released October 13 by Flashpoint, reveals that concerns over effectiveness and reliability keep the majority of security professionals from embracing autonomous AI agents. The report suggests that while companies pour more resources into agentic AI, skepticism about trusting these systems with critical decisions remains widespread.
Among the three-quarters of security teams not yet using agentic AI in full deployment, 20% are conducting closed pilot programs, while another 35% are evaluating potential adoption over the next 12 to 24 months. The remaining 20% of respondents have no current plans to test agentic AI for cyber defense. Organizations that are using or piloting AI agents assign them tasks including retrieving information and prioritizing alerts, as well as more hands-on activities such as blocking IP addresses, updating blacklists, and monitoring suspicious traffic. Some companies have gone further, deploying AI to take automated actions that directly influence security operations, though many remain hesitant about granting AI agents this level of decision-making power.
Cybersecurity professionals cited multiple obstacles to broader adoption, with erroneous AI decisions emerging as a primary worry. "An agent that's confidently wrong is worse than one that's slow," one respondent noted. Additional barriers include data leakage risks, auditability challenges, regulatory compliance requirements, cost considerations, and uncertainty about how much control agents should exercise over security workflows. According to Josh Lefkowitz, founder and CEO of Flashpoint, security leaders recognize the significant opportunity agentic AI presents on the defensive side and are prepared to invest, but they're not influenced by hype alone. He explained that these leaders are holding agentic AI to a rigorous standard, wanting to understand where it enhances their work, whether the output can be trusted, and where human judgment must stay in the loop.
Despite practitioner caution, boards and executive funding authorities appear enthusiastic about agentic AI adoption, with more than half anticipating substantial budget increases to support investigation of the technology for cyber defense. Over the next year, 35% expect their budgets to climb by more than 10%, while an additional 15% foresee increases up to 10%. No respondents predicted their cybersecurity budgets would shrink. The report indicates that while security teams continue testing agentic AI cautiously, the gap between executive enthusiasm and practitioner readiness highlights the tension between strategic investment and operational confidence. Organizations face pressure to modernize defenses against AI-enabled attacks while simultaneously grappling with fundamental questions about whether these tools can perform reliably when stakes are highest. The pressure to adopt cutting-edge defenses without sacrificing trust or control will likely define enterprise security strategy as autonomous systems mature. For now, cybersecurity leaders appear committed to rigorous validation before handing over the keys.