Sixty-nine percent of enterprise IT teams cannot reverse a faulty AI-generated device configuration within an hour, and 43% need more than 24 hours, according to a new report from Fleet Device Management.

The findings come from a Fleet survey of more than 250 enterprise IT practitioners who manage Apple devices. The full report, titled "Apple in the Enterprise," is available on Fleet's website. The data exposes a widening disconnect between AI-driven deployment speed and recovery capability inside macOS, iOS, and iPadOS management workflows.

What did the survey find?

Headline numbers from the report:

• 86% of respondents let AI-written output reach production devices after some review
• 61% used AI to author an MDM profile or configuration in the past month
• 62% used AI to write scripts or code
• 69% cannot roll back a bad configuration within an hour
• 43% need more than a day to recover
• 26% can recover the same day, but only with manual intervention

Fleet founder and CEO Mike McNeil framed the exposure bluntly. "When one of those changes is wrong, 69% can't undo it within an hour, and 43% need more than 24 hours," he said. "So the exposure is real even without a count of incidents. The AI tool isn't what gets pulled back. The configuration it produced is, and most teams do that by hand."

Why is MDM a high-value target?

McNeil separated risk into two axes: the privilege level of the code and the difficulty of recovery. Scripts running as root can seize the machine and leave no clean revert path.

He singled out Microsoft Intune as the most common Apple management platform among respondents, at 36%, despite Intune's Windows-first design. "Tooling built around one platform's assumptions tends to be weakest at the edges of another's," McNeil said.

He called MDM one of the few trusted channels that can bypass macOS platform-security prompts. "MDM is one of the few trusted paths that can bypass the [macOS platform security] prompts, which is exactly why a bad or malicious MDM-delivered change is so consequential," he warned.

How should IT close the rollback gap?

McNeil proposed treating profiles, scripts, and policies as application code, stored as files in a Git repository, with the Fleet MDM pipeline as one working example. His recommended pattern:

• Store device configuration as files in a Git repo
• Submit changes as pull requests
• Require a second-person review
• Apply changes through automation
• Revert by reverting the commit

He flagged one caveat: "A revert fixes what the configuration says. It doesn't undo a script that has already run. That's an argument for preferring declarative configuration over imperative scripts where you can."

What strains Apple's enterprise admins most?

Keeping pace with Apple's release cycle, not budget, tops the list. Thirty-one percent of respondents named Apple's release cadence as the single hardest part of the job. Budget and headcount trailed at 5%.

"Keeping up with Apple's release pace was the single most cited hardest part of the job, at 31%," McNeil said. "Admins aren't complaining that Apple ships too little. They're stretched by the yearly OS cycle, new frameworks, and deprecations."

He endorsed the direction of Apple's enterprise stack. "Declarative Device Management and tighter platform security are good for enterprises," McNeil said. "The pace is the issue, and it's the main reason admins need automation and a safety net."

Only 7.6% of Fleet's customers run Mac exclusively, which McNeil uses to argue that the AI-rollback gap is not Apple-specific. Windows and Linux carry the heaviest scripting loads and the largest script-error surface, in his view.

What does Fleet recommend going forward?

"Speed is only an advantage if recovery keeps pace," McNeil said. He urged admins to apply the same controls they would apply to any production change: review, least privilege, a change record, and a fast revert.

He summarized the new workload as "less clicking, more engineering." As AI drafts more profiles and custom scripts, McNeil expects admins to shift from making changes to deciding which changes to make.