Survey Finds AI Adoption in Security Operations Outpacing Integration
Summary
A Torq-commissioned Sapio Research survey of 450+ security leaders finds AI use in SOCs is fragmented: teams average seven AI solutions, while respondents cite trust and explainability concerns and favor unified platforms.
Key points
- Sapio Research conducted the survey on behalf of Torq, gathering responses from more than 450 security leaders at enterprises across four countries.
- The report says 79% of organizations use AI in at least one SOC function, 54% have incorporated it into workflows, and 37% have adopted it widely.
- Respondents’ SOCs use an average of seven AI solutions; 80% rely on multiple point tools, and 85% would prefer a unified platform.
- Some form of AI autonomy is acceptable to 97% of respondents; 72% are comfortable with fully autonomous AI for medium-severity alerts or below.
- The report identifies data privacy, missed threats, data governance, false positives, and loss of control among concerns eroding trust; 90% say explainability is important.
- Analysts reportedly spend 8.6 hours per week overseeing AI outputs, while nine in ten leaders say AI has positively affected SOC workload and reduced stress or burnout.
- For the next 12 months, 40% of respondents plan to expand AI use in cloud security and 28% in incident-response automation.
Article Details
- Publisher
- Sapio Research on behalf of Torq
- Report Period
- 2026 report; survey fieldwork dates not disclosed
- Scope
- AI adoption, use, confidence, and operational barriers in enterprise SOCs; respondents were security leaders at organizations with 1,000+ employees across four countries.
- Sample Size
- More than 450 CISOs, security leaders, and senior security decision-makers.
- Key Statistics
- Respondents reported an average of 7 AI solutions per SOC.
- 80% said their organizations rely on fragmented point tools.
- 85% preferred a unified AI SOC platform over multiple point solutions.
- 90% said explainable AI decisions are important for a true AI SOC.
- Analysts reported spending 8.6 hours per week overseeing AI-powered outputs.
- Recommendations
- Make AI decision-making transparent and explainable.
- Provide strong data privacy protections and governance.
- Use adjustable autonomy and human-in-the-loop controls calibrated to severity, confidence, and context.
- Integrate with existing security tools and enable consolidation without requiring rip-and-replace.
- Use analyst overrides as feedback to improve AI over time.
People
Vendors
Products
Countries
France- **France**: 53% Agentic | 45% Custom AI Agents | 42% Vendor AI AgentsGermany- **Germany**: *(Data point metrics distributed regionally in text)*United KingdomThe UK leads in broad deployment at 48%, while other markets show strong but less uniform penetration.United States- **U.S.**: *(Data point metrics distributed regionally in text)*