OpenAI ChatGPT Enterprise Gmail Message Sent via Connector


Description

Detects ChatGPT Enterprise Gmail connector send_email responses. Adversaries who abuse a linked Gmail connector after session compromise can exfiltrate data or perform lateral phishing by sending mail on behalf of the user. Draft-only activity (create_draft) is excluded.

Query · esql

from logs-openai_chatgpt_enterprise.app_log-* metadata _id, _version, _index
| where
    data_stream.dataset == "openai_chatgpt_enterprise.app_log" and
    openai_chatgpt_enterprise.app_log.app_name == "Gmail" and
    openai_chatgpt_enterprise.app_log.app_type == "SERVICE" and
    openai_chatgpt_enterprise.app_log.output is not null
| eval Esql.gmail_action = field_extract(openai_chatgpt_enterprise.app_log.output, "response.action_name")
| where Esql.gmail_action == "send_email"
| keep
    _id, _version, _index, @timestamp, user.id, user.email,
    Esql.gmail_action,
    openai_chatgpt_enterprise.app_log.app_name,
    openai_chatgpt_enterprise.app_log.app_type,
    openai_chatgpt_enterprise.app_log.log_type,
    openai_chatgpt_enterprise.app_log.output,
    data_stream.*

Investigation fields

Pivot points the source recommends for triage.

  • @timestamp
  • user.email
  • user.id
  • Esql.gmail_action
  • openai_chatgpt_enterprise.app_log.input
  • openai_chatgpt_enterprise.app_log.app_name

Implementation guide

This rule requires the OpenAI ChatGPT Enterprise integration to collect Gmail app_log response events where openai_chatgpt_enterprise.app_log.output includes response.action_name (for example send_email).

https://www.elastic.co/docs/reference/integrations/openai_chatgpt_enterprise

Known false positives

  • Users legitimately asking ChatGPT to send mail through an approved Gmail connector.

Analyst notes

Investigating OpenAI ChatGPT Enterprise Gmail Message Sent via Connector

ChatGPT completed a Gmail connector send_email call. That can be legitimate assistance or abuse for exfiltration or phishing after connector or session compromise.

Escalate when the recipient or content on the paired request is unexpected, the message looks like phishing or data staging, or nearby activity shows sensitive Gmail or Drive discovery. Close as a false positive for approved send-on-behalf workflows.

Possible investigation steps

  • Review user.email, Esql.gmail_action, and openai_chatgpt_enterprise.app_log.output (sent message and thread IDs).
  • Correlate with the preceding Gmail request from the same user (subject, body, and to are on the request event).
  • Correlate with nearby Gmail discovery searches, Google Drive downloads, or unusual authentication activity.
  • Confirm whether Gmail connector use and outbound send are approved for that user.

False positive analysis

  • Users who send calendar notes, status updates, or other approved outbound mail through the Gmail connector will match.

Response and remediation

  • Revoke Gmail connector access if unauthorized, review sent mail, and treat exposed content per your data-incident process.
  • Hunt for additional sends, sensitive Gmail searches, or Drive downloads from the same user in the surrounding window.
Raw source OpenAI ChatGPT Enterprise Gmail Message Sent via Connector · Elastic TOML
Esc
Published by elastic/detection-rules ↗, licensed under Elastic License 2.0 ↗. Reproduced here unmodified.
[metadata]
creation_date = "2026/10/09"
integration = ["openai_chatgpt_enterprise"]
maturity = "production"
min_stack_comments = "field_extract (flattened field support) is available starting in 9.5.0, as a Technical Preview function; GA in 9.6.0."
min_stack_version = "9.5.0"
updated_date = "2026/10/09"

[rule]
author = ["Elastic"]
description = """
Detects ChatGPT Enterprise Gmail connector `send_email` responses. Adversaries who abuse a linked Gmail connector after
session compromise can exfiltrate data or perform lateral phishing by sending mail on behalf of the user. Draft-only
activity (`create_draft`) is excluded.
"""
false_positives = ["Users legitimately asking ChatGPT to send mail through an approved Gmail connector."]
from = "now-9m"
language = "esql"
license = "Elastic License v2"
name = "OpenAI ChatGPT Enterprise Gmail Message Sent via Connector"
note = """## Triage and analysis

### Investigating OpenAI ChatGPT Enterprise Gmail Message Sent via Connector

ChatGPT completed a Gmail connector `send_email` call. That can be legitimate assistance or abuse for exfiltration or
phishing after connector or session compromise.

Escalate when the recipient or content on the paired request is unexpected, the message looks like phishing or data
staging, or nearby activity shows sensitive Gmail or Drive discovery. Close as a false positive for approved
send-on-behalf workflows.

#### Possible investigation steps

- Review `user.email`, `Esql.gmail_action`, and `openai_chatgpt_enterprise.app_log.output` (sent message and thread IDs).
- Correlate with the preceding Gmail `request` from the same user (subject, body, and `to` are on the request event).
- Correlate with nearby Gmail discovery searches, Google Drive downloads, or unusual authentication activity.
- Confirm whether Gmail connector use and outbound send are approved for that user.

### False positive analysis

- Users who send calendar notes, status updates, or other approved outbound mail through the Gmail connector will match.

### Response and remediation

- Revoke Gmail connector access if unauthorized, review sent mail, and treat exposed content per your data-incident
  process.
- Hunt for additional sends, sensitive Gmail searches, or Drive downloads from the same user in the surrounding window.
"""
references = [
    "https://www.elastic.co/docs/reference/integrations/openai_chatgpt_enterprise",
    "https://platform.openai.com/docs/compliance",
]
risk_score = 47
rule_id = "7184e35d-5605-4283-badf-435e1974e488"
setup = """## Setup

This rule requires the OpenAI ChatGPT Enterprise integration to collect Gmail `app_log` response events where
`openai_chatgpt_enterprise.app_log.output` includes `response.action_name` (for example `send_email`).

https://www.elastic.co/docs/reference/integrations/openai_chatgpt_enterprise
"""
severity = "medium"
tags = [
    "Domain: GenAI",
    "Platform: OpenAI",
    "Data Source: OpenAI ChatGPT Enterprise",
    "Service: OpenAI ChatGPT Enterprise",
    "Use Case: Threat Detection",
    "Resources: Investigation Guide",
    "Rule Type: ES|QL",
    "Tactic: Exfiltration",
    "Tactic: Lateral Movement",
    "Mitre Atlas: AML.T0053",
    "Mitre Atlas: AML.T0025",
]
timestamp_override = "event.ingested"
type = "esql"

query = '''
from logs-openai_chatgpt_enterprise.app_log-* metadata _id, _version, _index
| where
    data_stream.dataset == "openai_chatgpt_enterprise.app_log" and
    openai_chatgpt_enterprise.app_log.app_name == "Gmail" and
    openai_chatgpt_enterprise.app_log.app_type == "SERVICE" and
    openai_chatgpt_enterprise.app_log.output is not null
| eval Esql.gmail_action = field_extract(openai_chatgpt_enterprise.app_log.output, "response.action_name")
| where Esql.gmail_action == "send_email"
| keep
    _id, _version, _index, @timestamp, user.id, user.email,
    Esql.gmail_action,
    openai_chatgpt_enterprise.app_log.app_name,
    openai_chatgpt_enterprise.app_log.app_type,
    openai_chatgpt_enterprise.app_log.log_type,
    openai_chatgpt_enterprise.app_log.output,
    data_stream.*
'''


[[rule.threat]]
framework = "MITRE ATT&CK"
[[rule.threat.technique]]
id = "T1567"
name = "Exfiltration Over Web Service"
reference = "https://attack.mitre.org/techniques/T1567/"


[rule.threat.tactic]
id = "TA0010"
name = "Exfiltration"
reference = "https://attack.mitre.org/tactics/TA0010/"
[[rule.threat]]
framework = "MITRE ATT&CK"
[[rule.threat.technique]]
id = "T1534"
name = "Internal Spearphishing"
reference = "https://attack.mitre.org/techniques/T1534/"


[rule.threat.tactic]
id = "TA0008"
name = "Lateral Movement"
reference = "https://attack.mitre.org/tactics/TA0008/"
[[rule.threat_mappings]]
framework = "MITRE ATLAS"
version = "2026.08"
[[rule.threat_mappings.threat]]
framework = "MITRE ATLAS"
[[rule.threat_mappings.threat.technique]]
id = "AML.T0053"
name = "AI Agent Tool Invocation"
reference = "https://atlas.mitre.org/techniques/AML.T0053/"


[rule.threat_mappings.threat.tactic]
id = "AML.TA0005"
name = "Execution"
reference = "https://atlas.mitre.org/tactics/AML.TA0005/"
[[rule.threat_mappings.threat]]
framework = "MITRE ATLAS"
[[rule.threat_mappings.threat.technique]]
id = "AML.T0025"
name = "Exfiltration via Cyber Means"
reference = "https://atlas.mitre.org/techniques/AML.T0025/"


[rule.threat_mappings.threat.tactic]
id = "AML.TA0010"
name = "Exfiltration"
reference = "https://atlas.mitre.org/tactics/AML.TA0010/"

[rule.investigation_fields]
field_names = [
    "@timestamp",
    "user.email",
    "user.id",
    "Esql.gmail_action",
    "openai_chatgpt_enterprise.app_log.input",
    "openai_chatgpt_enterprise.app_log.app_name",
]

Detection rules belong to the projects that publish them and remain under their own licenses. This site indexes and links to them; it claims no rights in them.