How to Fix SentinelOne Static AI False Positives on macOS Tahoe 26.6

Overview

Following the macOS Tahoe 26.6 update released on July 27, 2026, SentinelOne environments running the Static AI detection engine began generating false positive alerts against legitimate, Apple-signed macOS system processes. The most commonly reported alert involves the mds system process — the Spotlight metadata server — which the Static AI engine misidentified as a variant of the "Protected" malware family.

The permanent fix is an agent update. In the meantime, an Alerts Exclusion will suppress the noise, but only if it is explicitly scoped to the Static AI engine — a file path exclusion alone will not stop the alert recurring.

This article covers the symptom, why the obvious exclusion fails, the interim suppression procedure, and the agent versions that resolve it permanently.

This is a false positive. The detection does not indicate compromise or an active threat on the endpoint.

Symptom

After updating affected endpoints to macOS Tahoe 26.6, administrators observe repeated Low-severity alerts in the SentinelOne Management Console with the following characteristics:

Because mds runs as root and is spawned repeatedly by launchd as part of normal macOS operation, affected endpoints can generate a high volume of repeated alerts.

Root Cause

SentinelOne has identified that the macOS Tahoe 26.6 update altered attributes of certain signed Apple system processes in a way that caused the Static AI engine's on-write classification model to misclassify them as a variant of the "Protected" malware family.

This is a detection-logic issue in the Static AI signature set, not an indication of compromise.

Why a File Path Exclusion Alone Doesn't Resolve It

Administrators who create an Alerts Exclusion scoped only to the file path — without explicitly scoping the exclusion to the Static AI detection engine — will continue to see the alert recur, even when the exclusion is applied at the account level.

The default exclusion configuration does not automatically suppress detections from every engine. The exclusion must be explicitly applied against the same detection engine that generated the alert.

This is the single most common reason administrators report that the exclusion "isn't working."

Interim Fix: Create a Correctly Scoped Alerts Exclusion

Use this to stop the alert while the permanent fix is applied.

Step 1: Confirm the detection engine

Open the alert and confirm the Detection Engine field shows On-Write Static AI.

Step 2: Open the exclusion

Go to Edit Alerts Exclusion, or create a new Alerts Exclusion.

Step 3: Set the exclusion type and path

Set the exclusion type to File Path and enter:

/System/Library/Frameworks/CoreServices.framework/Versions/A/Frameworks/Metadata.framework/Versions/A/Support/mds

Step 4: Scope the exclusion to Static AI

Under Detection Engines, select Specific Engines and check Static AI.

This step is what makes the exclusion work. Leaving the default all-engines behaviour in place is why path-only exclusions continue to alert.

Step 5: Set scope and save

Set the appropriate scope — Account, Site, or Group — and save.

This interim exclusion suppresses the false alert but does not address the underlying detection logic issue.

Permanent Resolution

SentinelOne has released updates that correct the Static AI detection logic.

Live Security Update (LSU). A phased LSU addressing this issue has now been fully rolled out to eligible environments with Live Security Updates enabled. No manual action is required once the update has propagated to an endpoint.

Service Packs (for environments not using LSUs). For macOS agent versions 25.4 and 26.1, dedicated service packs are available:

Recommendation: upgrade affected macOS agents to version 25.4.3 SP1 or 26.1.3 SP1.

If you manage macOS agent versions through Intune rather than the SentinelOne console, see the guide on deploying the SentinelOne macOS agent using Intune for the packaging and configuration profile workflow.

Validation

Confirm the fix has taken effect before considering the issue closed:

• The affected endpoint reports the updated agent version — 25.4.3 SP1 or 26.1.3 SP1 — in the console, or has received the LSU

• No new mds Static AI detections appear against the endpoint after the update propagates

• Existing alerts can be resolved as false positive

Once the update or service pack has been applied, the interim Alerts Exclusion created above can be safely removed. Leaving it in place indefinitely suppresses a legitimate detection path on a root-level system process.

Common Issues and Troubleshooting

Exclusion is saved but the alert keeps recurring

The exclusion was not scoped to a specific detection engine. Reopen it, select Specific Engines, and check Static AI. A file path exclusion with default engine scope will not suppress a Static AI detection.

Alert still appears after applying the exclusion at account level

Scope is not the issue — engine selection is. Account-level scope does not compensate for an unspecified detection engine.

Endpoint has not received the Live Security Update

Confirm Live Security Updates are enabled for the environment. Where they are not, apply the applicable service pack manually rather than waiting for propagation.

Similar detections on other Apple-signed system processes

The root cause affects attributes of signed Apple system processes generally, so mds may not be the only process flagged in a given environment. Verify the signature status and detection engine on each alert before excluding, and prioritise the agent update over a growing list of per-path exclusions.

Security Considerations

Exclusions on root-level system paths reduce visibility, so treat this one as temporary by design. Record it with an owner and a removal trigger — the agent upgrade — rather than leaving it to accumulate as permanent configuration. For a case where exclusions are the intended long-term answer rather than a stopgap, see resolving SentinelOne high CPU usage on developer machines.

Before excluding, confirm the alert genuinely matches the profile above: Signed & Verified with Apple Inc. as publisher, On-Write Static AI as the engine, and the exact system path. A detection on an unsigned binary, or on a similarly named process at a different path, is not this issue and should be investigated rather than suppressed.

Applying the agent update remains the correct long-term action. It restores full Static AI coverage on the affected processes instead of leaving a permanent blind spot behind.

Managing SentinelOne false positives, exclusion policy, or agent update rollout across a macOS fleet? Talk to NetNXT →

FAQs

1) Why is SentinelOne flagging the macOS mds process as malware?

The macOS Tahoe 26.6 update altered attributes of signed Apple system processes, causing the Static AI engine to misclassify mds as a variant of the "Protected" malware family. It is a detection-logic issue, not a compromise.

2) Why isn't my SentinelOne file path exclusion stopping the alert?

The exclusion must be scoped to the detection engine that generated the alert. Under Detection Engines, select Specific Engines and check Static AI — path alone is not sufficient.

3) Which SentinelOne agent version fixes the macOS Tahoe 26.6 false positive?

Upgrade to macOS agent version 25.4.3 SP1 (25.4.3.8807) or 26.1.3 SP1 (26.1.3.8805). Environments with Live Security Updates enabled receive the fix automatically.

4) Should I remove the exclusion after updating the SentinelOne agent?

Yes. Once the LSU or service pack has been applied, remove the interim exclusion so Static AI coverage is restored on the affected system path.

5) Does the SentinelOne mds detection indicate a real threat?

No. The process is Apple-signed and verified, with a 0% true positive community verdict. The alert is a false positive caused by the Static AI signature set.