Most security teams now use AI, but few check if they're ready

Forty percent of security teams now use AI every day, and another 56% are testing it. Only 4% have no plans to adopt it. Those figures come from Prophet Security's State of AI in Security Operations 2026 survey of more than 250 security leaders and practitioners. The company now argues that teams still testing should ask a harder question: is our setup actually ready?

What the survey found

A SOC, or security operations center, is the team that watches for threats and investigates alerts. Respondents said 28% of their alerts are never investigated. A thorough investigation takes an average of 75 minutes, and alerts often wait close to an hour before anyone opens them. AI agents are built to start on each alert as it arrives, at any hour.

What readiness means

The article says readiness has two parts: having the problems AI agents are built for, and having the groundwork they need. A team does not need to be far along a maturity model to qualify. The groundwork covers four areas:

• Access to the tools and data an analyst would use, so the agent can see the full picture.
• A written standard for how alerts are investigated, plus baseline numbers to compare against.
• Rules for what the agent may do on its own.
• A plan for how to use the analyst hours the agent gives back.

Trust and privacy

Teams remain careful. In the survey, 57% require a person to review every AI decision before an alert is closed. On response, 44% have AI recommend actions that people carry out, and 30% let it handle low-risk fixes alone. No respondent gives it full, unsupervised control. The top roadblock, named by 44%, was data privacy and how AI models are trained.

What to do

The article suggests starting with read-only access and widening permissions as confidence grows, as with a new human analyst. A simple test for gaps: take five recent incidents, list every query a senior analyst ran, and note any the agent could not run. Buyers should also ask each vendor whether their data trains its models and where it is processed. Note that the survey and framework come from Prophet Security, which sells AI security tools.