Companies have invested more resources than ever into agentic AI as a cyber defense tool amid growing AI-enabled cyber threats to enterprise networks from. However, the majority of cybersecurity teams remain reluctant to deploy the technology due to concerns over their effectiveness.
Published on October 13, The Enterprise Threat Intelligence Report: 2026 from Flashpoint suggested that despite the ongoing discussion surrounding the potential cyber risks around offensive agentic AI, just a quarter (25%) of cybersecurity teams are actively using agentic AI tools as part of their active cyber defense strategy.
While three quarters of security teams have yet to fully investigate the potential deployment of AI tools into their security operations, 20% of those surveyed said that they are running a closed pilot scheme, and a further 35% said they are examining the potential of adoption within the next 12 to 24 months. The remaining 20% of respondents said they have no plans to test out agentic AI as part of their cyber defense as things stand.
Of those who are using or are piloting AI agents, tasks these tools are involved in include information retrieval and alert prioritization, as well as more active tasks such as blocking IP addresses, updating blacklists and monitoring suspicious traffic.
Some organizations have also used AI to help with more advanced activity, including taking automated action that can directly affect security operations. However, many remain wary of providing an AI agent with this amount of decision-making authority.
When AI Agents Are ‘Confidently Wrong’
Some cybersecurity professionals expressed concerns about agents making wrong decisions.
“An agent that’s confidently wrong is worse than one that’s slow,” said one respondent.
Other hurdles to deployment of AI agents in cyber defense include concerns about data leakage, auditability, regulatory requirements and cost, as well as the level of control agents should have over security workflows.
While many organizations are toying with the investment in and deployment of agentic AI as a cyber defense tool, many are still reluctant to go all in until they have done more research into whether AI agents can be trusted with tasks.
“The opportunity for agentic AI on the defensive side is significant, and our research shows security leaders are ready to invest. But they aren't swayed by the hype,” said Josh Lefkowitz, founder and CEO of Flashpoint.
“They're holding agentic AI to a high bar, and they want to know where it improves their work, whether they can trust the output, and where human judgment must remain part of the process."
While cybersecurity teams are still testing the waters with agentic AI, the report also suggests that boards and executive funding providers appear keen on the adoption of the technology, with over half expecting a significant budget increase to aid with the investigation of agentic AI for cyber defense.
More than a third (35%) said that over the next 12 months, they expect their budget to rise by over 10%, while a further 15% suggested they expect to see an increase of up to 10%. Zero respondents said they believed their cybersecurity budget is set to decrease.