Detection: M365 Copilot Agentic Jailbreak Attack

EXPERIMENTAL DETECTION

This detection status is set to experimental. The Splunk Threat Research team has not yet fully tested, simulated, or built comprehensive datasets for this detection. As such, this analytic is not officially supported. If you have any questions or concerns, please reach out to us at research@splunk.com.

Updated Date: 2025-09-25 ID: e5c7b380-19da-42e9-9e53-0af4cd27aee3 Author: Rod Soto Type: Anomaly Product: Splunk Enterprise Security

Description

Detects agentic AI jailbreak attempts that try to establish persistent control over M365 Copilot through rule injection, universal triggers, response automation, system overrides, and persona establishment techniques. The detection analyzes the PromptText field for keywords like "from now on," "always respond," "ignore previous," "new rule," "override," and role-playing commands (e.g., "act as," "you are now") that attempt to inject persistent instructions. The search computes risk by counting distinct jailbreak indicators per user session, flagging coordinated manipulation attempts.

Search

 1`m365_exported_ediscovery_prompt_logs` 
 2| eval user = Sender 
 3| eval rule_injection=if(match(Subject_Title, "(?i)(rules
 4|instructions)\s*="), "YES", "NO") 
 5| eval universal_trigger=if(match(Subject_Title, "(?i)(every
 6|all).*prompt"), "YES", "NO") 
 7| eval response_automation=if(match(Subject_Title, "(?i)(always
 8|automatic).*respond"), "YES", "NO") 
 9| eval system_override=if(match(Subject_Title, "(?i)(override
10|bypass
11|ignore).*(system
12|default)"), "YES", "NO") 
13| eval persona_establishment=if(match(Subject_Title, "(?i)(with.*\[.*\]
14|persona)"), "YES", "NO") 
15| where rule_injection="YES" OR universal_trigger="YES" OR response_automation="YES" OR system_override="YES" OR persona_establishment="YES" 
16| table _time, "Source ID", user, Subject_Title, rule_injection, universal_trigger, response_automation, system_override, persona_establishment, Workload 
17| sort -_time 
18| `m365_copilot_agentic_jailbreak_attack_filter`

Data Source

Name	Platform	Sourcetype	Source
M365 Exported eDiscovery Prompts	N/A	`'csv'`	`'csv'`

Macros Used

Name	Value
m365_exported_ediscovery_prompt_logs	`(sourcetype=csv)`
m365_copilot_agentic_jailbreak_attack_filter	`search *`

m365_copilot_agentic_jailbreak_attack_filter is an empty macro by default. It allows the user to filter out any results (false positives) without editing the SPL.

Annotations

- MITRE ATT&CK

+ Kill Chain Phases

+ NIST

+ CIS

- Threat Actors

ID	Technique	Tactic
T1562	Impair Defenses	Defense Evasion

Exploitation

DE.AE

CIS 10

Magic Hound

Default Configuration

This detection is configured by default in Splunk Enterprise Security to run with the following settings:

Setting	Value
Disabled	true
Cron Schedule	`0 * * * *`
Earliest Time	`-70m@m`
Latest Time	`-10m@m`
Schedule Window	`auto`
Creates Risk Event	True

This configuration file applies to all detections of type anomaly. These detections will use Risk Based Alerting.

Implementation

To export M365 Copilot prompt logs, navigate to the Microsoft Purview compliance portal (compliance.microsoft.com) and access eDiscovery. Create a new eDiscovery case, add target user accounts or date ranges as data sources, then create a search query targeting M365 Copilot interactions across relevant workloads. Once the search completes, export the results to generate a package containing prompt logs with fields like Subject_Title (prompt text), Sender, timestamps, and workload metadata. Download the exported files using the eDiscovery Export Tool and ingest them into Splunk for security analysis and detection of jailbreak attempts, data exfiltration requests, and policy violations.

Known False Positives

Legitimate users discussing AI ethics research, security professionals testing system robustness, developers creating training materials for AI safety, or academic discussions about AI limitations and behavioral constraints may trigger false positives.

Associated Analytic Story

Suspicious Microsoft 365 Copilot Activities

Risk Based Analytics (RBA)

Risk Message:

User $user$ attempted to establish persistent agentic control over M365 Copilot through advanced jailbreak techniques including rule injection, universal triggers, and system overrides, potentially compromising AI security across multiple sessions.

Risk Object	Risk Object Type	Risk Score	Threat Objects
user	user	50	No Threat Objects

References

https://www.splunk.com/en_us/blog/artificial-intelligence/m365-copilot-log-analysis-splunk.html

Detection Testing

Test Type	Status	Dataset	Source	Sourcetype
Validation	Not Applicable	N/A	N/A	N/A
Unit	✅ Passing	Dataset	`csv`	`csv`
Integration	✅ Passing	Dataset	`csv`	`csv`

Replay any dataset to Splunk Enterprise by using our replay.py tool or the UI. Alternatively you can replay a dataset into a Splunk Attack Range

Source: GitHub | Version: 1