Detection rules › Splunk

Windows Process Accessing Windows Recall Directory

Status
production
Severity
low
Group by
access_mask, computer_name, object_name, process_name, signature_id
Author
Raven Tait, Splunk
Source
github.com/splunk/security_content

This detection triggers on a process accessing the Windows Recall directory. Recall is a new feature Microsoft release that takes screenshots every 5 or so seconds to provide context to it's AI features. The initial release of Recall was lacking in the security department due to it being trivial to view and steal the data. Due to this lack of security it's likely that info stealer malware will take advantage of this new feature. Microsoft has recognized the security issues with Recall and is planning on making improvements. Once those improvements are released we will re-assess this detection to make sure it is still relevant.

MITRE ATT&CK coverage

Event coverage

Rule body splunk

name: Windows Process Accessing Windows Recall Directory
id: d1d7048e-d095-4fb5-b43b-e570d574a1aa
version: 2
creation_date: '2026-05-05'
modification_date: '2026-05-13'
author: Raven Tait, Splunk
status: production
type: Anomaly
description: |-
    This detection triggers on a process accessing the Windows Recall directory.
    Recall is a new feature Microsoft release that takes screenshots every 5 or so seconds to provide context to it's AI features.
    The initial release of Recall was lacking in the security department due to it being trivial to view and steal the data.
    Due to this lack of security it's likely that info stealer malware will take advantage of this new feature.
    Microsoft has recognized the security issues with Recall and is planning on making improvements.
    Once those improvements are released we will re-assess this detection to make sure it is still relevant.
data_source:
    - Windows Event Log Security 4663
search: |-
    `wineventlog_security`
    AccessList="%%4416"
    EventID="4663"
    ObjectName="*CoreAIPlatform.00\\UKP*"
    NOT (
        ProcessName IN (
            "*aixhost.exe",
            "*aihost.exe"
        )
    )
    | fillnull
    | stats count min(_time) as firstTime
                  max(_time) as lastTime
      by Computer EventID AccessList ObjectName ProcessName
    
    | rename Computer as dest
    | `security_content_ctime(firstTime)`
    | `security_content_ctime(lastTime)`
    | `windows_process_accessing_windows_recall_directory_filter`
how_to_implement: The detection is based on data that originates from Endpoint Detection and Response (EDR) agents. These agents are designed to provide security-related telemetry from the endpoints where the agent is installed. To implement this search, you must ingest logs that contain the process GUID, process name, and parent process. Additionally, you must ingest complete command-line executions. These logs must be processed using the appropriate Splunk Technology Add-ons that are specific to the EDR product. The logs must also be mapped to the `Processes` node of the `Endpoint` data model. Use the Splunk Common Information Model (CIM) to normalize the field names and speed up the data modeling process.
known_false_positives: Legitimate system processes and trusted applications may access the Windows Recall directory for AI-related features. Review and allow access for approved software as needed.
references:
    - https://doublepulsar.com/recall-stealing-everything-youve-ever-typed-or-viewed-on-your-own-windows-pc-is-now-possible-da3e12e9465e
    - https://github.com/xaitax/TotalRecall
drilldown_searches:
    - earliest_offset: $info_min_time$
      latest_offset: $info_max_time$
      name: View the detection results for - "$user$" and "$dest$"
      search: '%original_detection_search% | search  user = "$user$" dest = "$dest$"'
    - name: View risk events for the last 7 days for - "$user$" and "$dest$"
      search: '| from datamodel Risk.All_Risk | search normalized_risk_object IN ("$user$", "$dest$") | stats count min(_time) as firstTime max(_time) as lastTime values(search_name) as "Search Name" values(risk_message) as "Risk Message" values(analyticstories) as "Analytic Stories" values(annotations._all) as "Annotations" values(annotations.mitre_attack.mitre_tactic) as "ATT&CK Tactics" by normalized_risk_object | `security_content_ctime(firstTime)` | `security_content_ctime(lastTime)`'
      earliest_offset: 7d
      latest_offset: "0"
intermediate_findings:
    entities:
        - field: dest
          type: system
          score: 20
          message: Potential Process Accessing Windows Recall Directory activity observed on $dest$.
analytic_story:
    - Windows Post-Exploitation
asset_type: Endpoint
mitre_attack_id:
    - T1119
    - T1059
product:
    - Splunk Enterprise
    - Splunk Enterprise Security
    - Splunk Cloud
category: endpoint
security_domain: endpoint
tests:
    - name: True Positive Test
      attack_data:
        - data: https://media.githubusercontent.com/media/splunk/attack_data/master/datasets/attack_techniques/T1119/snapattack/snapattack.log
          source: XmlWinEventLog:Security
          sourcetype: XmlWinEventLog
      test_type: unit

Stages and Predicates

Stage 1: search

`wineventlog_security`
AccessList="%%4416"
EventID="4663"
ObjectName="*CoreAIPlatform.00\\UKP*"
NOT (
    ProcessName IN (
        "*aixhost.exe",
        "*aihost.exe"
    )
)

Stage 2: fillnull

| fillnull

Stage 3: stats

| stats count min(_time) as firstTime
              max(_time) as lastTime
  by Computer EventID AccessList ObjectName ProcessName

Stage 4: rename

| rename Computer as dest

Stage 5: search

| `security_content_ctime(firstTime)`

Stage 6: search

| `security_content_ctime(lastTime)`

Stage 7: search

| `windows_process_accessing_windows_recall_directory_filter`

Exclusions

Top-level NOT(...) conjuncts: predicates this rule actively suppresses.

FieldKindExcluded values
ProcessNamein"*aihost.exe", "*aixhost.exe"

Indicators

Each row is a field, operator, and value that the rule matches. The corpus column counts how many other rules in the catalog look for the same combination: high numbers point to widely-used, community-vetted indicators. Blank or 1 shows that the indicator is specific to this rule.

FieldKindValues
AccessListeq
  • "%%4416" corpus 2 (splunk 2)
EventIDeq
  • "4663" corpus 34 (splunk 29, kusto 5)
ObjectNameeq
  • "*CoreAIPlatform.00\\UKP*"