AI adoption by sysadmins has not lived up to the optimism that marked early‑2024 forecasts, according to a recent 2026 survey from Action1.
Survey shows limited use of AI for critical tasks
The report examined expectations set two years earlier, when administrators anticipated that artificial intelligence would streamline patch management, vulnerability prioritization, infrastructure monitoring and incident response. In reality, fewer than 20 % of respondents reported deploying AI for either patch management or vulnerability prioritization.
Troubleshooting emerged as the only area where AI adoption approached the predicted level. Sysadmins said they rely on AI to generate hypotheses, pinpoint possible causes, locate relevant documentation and suggest next steps. Yet they still verify each output, redirect investigations when the model strays, and retain final responsibility for diagnosis and remediation.
Other functions, such as compliance analysis, first‑level end‑user support and post‑incident reviews, saw modest growth in AI use. Respondents noted recurring problems: outdated instructions, inaccurate recommendations, and difficulty handling user‑specific or environment‑specific issues.
Related: Traefik Labs launches secure runtime platform
Operational and security functions lag behind
Implementation changed little over the past two years in incident detection, vulnerability prioritization and patch management optimization. These tasks involve large volumes of data but require contextual understanding—system criticality, maintenance windows, compensating controls, business impact, dependencies and organizational risk tolerance—that AI currently lacks.
Policy development also remains a challenge. Effective AI assistance must account for internal governance, compliance obligations, risk appetite, ownership, existing procedures and organization‑specific exceptions. Without that context, recommendations risk misalignment with corporate requirements.
Software installation and maintenance present further hurdles. Errors such as selecting incorrect versions, overlooking dependencies, using invalid paths, generating nonexistent commands or providing instructions that do not match the target operating system can disrupt users, weaken security and affect production systems.
File management, user administration and identity management ranked among the least adopted use cases. Respondents reported incidents where access was incorrectly granted or revoked, credentials were exposed, permissions assigned to the wrong user, and files misclassified, moved or deleted. Such mistakes raise security, privacy and compliance concerns while leaving accountability ambiguous.
Related: CDW Cuts Jobs As AI Cost Cutting Deepens
While the share of companies mandating AI use by sysadmins more than doubled since 2023, adoption remains cautious. In 2026, 23 % of surveyed professionals said they had never used AI in a professional capacity. Data privacy, security risks and uncertainty about AI accuracy dominate concerns.
Trust levels vary by task. Sysadmins showed the highest confidence in AI for prioritization and scheduling. Nearly half trusted AI to assess the severity of newly released patches, schedule workstation patching, plan maintenance windows and reduce unnecessary after‑hours alerts. Conversely, they were reluctant to let AI make autonomous decisions affecting production systems. Most respondents demanded human approval for patching production servers, deferring patches, choosing rollbacks after failed updates and creating pilot groups for staged deployments.
When asked about unsupervised AI actions, 53 % would not allow AI to deploy patches across production systems while they were away, and only 14 % would consider it without supervision. Just 11 % would permit AI to override existing patching policies; 40 % would never allow such overrides.
AI remains a tool, not a master.
