Gartner: 60% of CISOs are piloting GenAI, but only 20% see results

The global threatscape is becoming dominated by all forms of weaponized LLMs, AI, and conversational agents, all aimed at launching lethal attacks that cripple companies and entire supply chains in minutes.

Nation‑state actors and organized eCrime groups now use artificial intelligence, including generative AI (GenAI), to automate reconnaissance, weaponize access, and strike faster than most defenses can respond. To keep pace, enterprises and the CISOs leading them are turning to GenAI as a defensive multiplier.

 CISOs are remaining optimistic

Gartner’s latest research quantifies that adoption is accelerating, but measurable results remain elusive. Approximately 60 % of organizations are piloting or planning GenAI cybersecurity initiatives. Only 20% of security leaders say these programs have delivered beneficial outcomes so far. These figures are from the research firm’s recent research note, What GenAI Use Cases Are Organizations Pursuing Within Cybersecurity? published earlier this month. Forrester predicts that the first agentic AI breach will happen in 2026.

Yet, despite early hurdles, cybersecurity leaders remain optimistic. Nearly every CISO I’ve spoken with sees GenAI as pivotal for transforming threat detection, proactive hunting, rapid incident response, and extracting actionable insights from terabytes of telemetry data streaming from endpoints and events. They recognize GenAI as crucial to decoding adversary tradecraft, particularly as identity-based threats and weaponized machine-learning attacks accelerate, reshaping the global threatscape in real time.

Key takeaways

• Code Analysis leads the pack. GenAI‑assisted code analysis is the most mature use case: 22% of enterprises use it today, and another 30% are piloting it. It addresses a persistent gap, as 69% of software‑engineering leaders cite insecure code remediation as a critical skills bottleneck.
• GenAI shows potential in helping SOC teams spot vulnerabilities faster. Currently, 21% of organizations actively leverage GenAI to enhance vulnerability detection and remediation, with another 26% piloting these capabilities. Adoption is driven by GenAI’s ability to automate vulnerability identification and prioritize remediation workflows, addressing longstanding security bottlenecks and resource constraints. Despite intense interest, widespread implementation remains challenged by integration complexity and skepticism about AI-generated accuracy, emphasizing the need for incremental deployment aligned with existing cybersecurity metrics.
• CISOs Shift from Ambition to Execution Gartner finds that the leaders gaining traction are those adopting “bite‑sized” implementations or use cases that fit into current processes, deliver quantifiable ROI, and build trust among analysts and engineers.

CISOs are dealing with a threatscape moving at machine speed

Given how lethal machine-driven attacks are becoming, exacerbated by the growing sophistication of weaponized AI, going on the offensive with GenAI is a choice more CISOs are considering.

• Nearly every cybersecurity team wants to have a Gen AI pilot either complete or in process to see how it integrates with their planned arsenal for 2026. Most CISOs want some form of AI in their arsenals going into the new year, as many expect the intensity, ingenuity, and lethal impact of automated attacks will reach new levels next year. One told me confidentially she fully expects machine-on-machine breach attempts to grow six times over in 2026 as her financial services firm handles highly speculative assets, including cryptocurrency ETFs and investment products.
• Breakout speed hits critical mass. CrowdStrike’s 2025 Global Threat Report reveals the alarming acceleration of attacks: the fastest observed eCrime intrusion took just 51 seconds to escalate from initial access to lateral movement, virtually eliminating defenders’ window to respond.
• Living-off-the-Land tactics dominate and often evade legacy cyberdefense systems: Malware-free intrusions surged significantly, now comprising 81% of interactive attacks in 2025. This trend is corroborated by findings from Mandiant and IBM X-Force, indicating adversaries are bypassing traditional signature-based controls by exploiting legitimate tools native to the enterprise environment.
• Nation-state activity reaching new record levels as weaponized tradecraft gains stealth and sophistication: CrowdStrike, Mandiant have documented triple-digit increases in operations linked to China, Iran, and North Korea. These attacks predominantly target telecommunications and critical infrastructure, reflecting geopolitical tensions and nation-states’ strategic prioritization of cyber-espionage.
• Global threat consensus is clear and compelling: ENISA’s Threat Landscape 2025 report aligns precisely with intelligence from CrowdStrike, Mandiant, and IBM X-Force, verifying that nation-state actors now leverage AI-driven automation to execute attacks faster than enterprises can detect, let alone defend.

CrowdStrike Founder and CEO George Kurtz underscored the urgency clearly in a recent CNBC interview on October 23rd, stating, “Well, this is something that we’ve really been focused on for the last number of years is being able to protect agentic AI. And if you think about agentic AI, it has the capabilities to interact with data. It has the capabilities to interact with Compute. It has identities, non-human identities, but it operates at superhuman speed. So all of the challenges that we’ve seen over the many years of humans getting themselves into trouble is only going to be exasperated by agentic AI, and we need security like CrowdStrike is delivering to protect it”.

Practical guidance from CISOs adding GenAI to their arsenals

Gartner’s latest research, combined with interviews and discussions with CISOs, security leaders, and SOC leaders who are piloting and in some cases using GenAI-based platforms today, offers this advice:

• Go deep on integration on pilots to see how strong the GenAI solution is as a contributor to your security tech stack: CISOs and SOC leaders tell me that this is the most reliable test of whether a GenAI platform or app will make the cut and get to production on their tech stack. Solid APIs that have been battle-tested by vendors who have a strong API management history have the inside track.
• Outcome-driven use cases are a must-have:At its core, cybersecurity is a business decision. And in a digital-first world, protecting your brand is essential. Any Gen AI pilot needs to contribute to a use case that makes a solid contribution to solidifying a business’s ability to compete.
• Start with time-tested, established metrics: Getting to a level of trust in GenAI is core to seeing if it is ready to progress from pilot into production. Evaluating GenAI effectiveness using established KPIs, including mean time to detect (MTTD) and mean time to respond (MTTR), at table stakes. CISOs and others running pilots caution about creating entirely new metrics just for GenAI. It obfuscates the total business impact of the technology.
• Parallel human trust and governance: Gartner emphasizes investing in employee enablement and robust governance frameworks like NIST’s AI Risk Management Framework to foster confidence in GenAI adoption. Human oversight remains a vital layer of control. Human-in-the-middle is essential for any workflow.

Bottom Line

Nation-state adversaries measure their innovation in how lethal their attacks are, how stealth their tradecraft is, and how easily they can evade legacy security techniques. It’s a full cyberwar just a few steps away from a full-on kinetic war. Research from CrowdStrike, IBM, Mandiant, and many other companies shows machine-to-machine attacks orchestrated with Gen AI are accelerating, so much so that Forrester predicts an imminent AI breach next year. GenAI’s ability to identify new threats and stop them makes the technology work a look.