Anthropic High File Upload Activity


Description

Detects an unusually high volume of Claude file uploads from the same user email and source IP within a rolling 24-hour period. Sustained upload activity can indicate staging of sensitive documents in Claude chats for later retrieval, automated ingestion of data into LLM workflows, or abuse of organizational Claude access to move files into the cloud service.

Query · esql

from logs-anthropic.audit-*
| where
    data_stream.dataset == "anthropic.audit" and
    event.action == "claude_file_uploaded" and
    event.outcome == "success" and
    user.email is not null and
    source.ip is not null
| stats
    Esql.event_count = count(*),
    Esql.event_id_values = values(event.id),
    Esql.file_name_values = values(file.name),
    Esql.anthropic_audit_claude_file_id_values = values(anthropic.audit.claude_file_id),
    Esql.anthropic_audit_claude_chat_id_values = values(anthropic.audit.claude_chat_id),
    Esql.anthropic_audit_claude_project_id_values = values(anthropic.audit.claude_project_id),
    Esql.user_agent_original_values = values(user_agent.original),
    Esql.anthropic_audit_actor_type_values = values(anthropic.audit.actor.type),
    Esql.timestamp_first_seen = min(@timestamp),
    Esql.timestamp_last_seen = max(@timestamp)
  by user.email, source.ip
| where Esql.event_count >= 100
| keep user.email, source.ip, Esql.*

Investigation fields

Pivot points the source recommends for triage.

  • user.email
  • source.ip
  • Esql.event_count
  • Esql.event_id_values
  • Esql.file_name_values
  • Esql.anthropic_audit_claude_file_id_values
  • Esql.anthropic_audit_claude_chat_id_values
  • Esql.anthropic_audit_claude_project_id_values
  • Esql.user_agent_original_values
  • Esql.anthropic_audit_actor_type_values
  • Esql.timestamp_first_seen
  • Esql.timestamp_last_seen

Known false positives

  • Developers, analysts, or knowledge workers bulk-uploading documents during migrations, evaluations, or large project onboarding can exceed the daily threshold from a single workstation or VPN egress.

Analyst notes

Investigating Anthropic High File Upload Activity

One user.email + source.ip pair uploaded a large volume of Claude files in 24 hours — possible staging for later retrieval, automated ingestion, or abuse of org Claude access.

Unauthorized / escalate when filenames look sensitive, uploads target shared/sensitive projects, UA looks automated (curl/python vs browser), or artifact sharing / exports follow. Close as FP for documented migrations, RAG pilots, or batch attach jobs with expected project IDs.

Possible investigation steps

  • Start with Esql.event_count, Esql.file_name_values, and chat/project ID lists — sensitive names or shared project IDs raise priority.
  • Triage user_agent on raw uploads: browser-like vs scripting clients. Scripting UAs with many distinct file IDs look more like automation than interactive work.
  • Pivot raw logs-anthropic.audit-* for individual anthropic.audit.claude_file_id / filenames and whether claude_file_viewed or chat activity matches the upload volume.
  • Correlate with artifact sharing, data exports, or compliance key creation from the same user.

False positive analysis

  • Document migrations and RAG indexing from one seat commonly exceed the threshold.

Response and remediation

  • On unauthorized staging: revoke sessions, review uploaded names/file IDs for sensitive content, and tighten project sharing or upload policy for the actor.
Raw source Anthropic High File Upload Activity · Elastic TOML
Esc
Published by elastic/detection-rules ↗, licensed under Elastic License 2.0 ↗. Reproduced here unmodified.
[metadata]
creation_date = "2026/09/15"
integration = ["anthropic"]
maturity = "production"
updated_date = "2026/09/25"

[rule]
author = ["Elastic"]
description = """
Detects an unusually high volume of Claude file uploads from the same user email and source IP within a rolling 24-hour
period. Sustained upload activity can indicate staging of sensitive documents in Claude chats for later retrieval,
automated ingestion of data into LLM workflows, or abuse of organizational Claude access to move files into the cloud
service.
"""
false_positives = [
    """
    Developers, analysts, or knowledge workers bulk-uploading documents during migrations, evaluations, or large project
    onboarding can exceed the daily threshold from a single workstation or VPN egress.
    """,
]
from = "now-24h"
interval = "1h"
language = "esql"
license = "Elastic License v2"
name = "Anthropic High File Upload Activity"
note = """## Triage and analysis

### Investigating Anthropic High File Upload Activity

One `user.email` + `source.ip` pair uploaded a large volume of Claude files in 24 hours — possible staging for later
retrieval, automated ingestion, or abuse of org Claude access.

Unauthorized / escalate when filenames look sensitive, uploads target shared/sensitive projects, UA looks automated
(curl/python vs browser), or artifact sharing / exports follow. Close as FP for documented migrations, RAG pilots, or
batch attach jobs with expected project IDs.

#### Possible investigation steps

- Start with `Esql.event_count`, `Esql.file_name_values`, and chat/project ID lists — sensitive names or shared
  project IDs raise priority.
- Triage `user_agent` on raw uploads: browser-like vs scripting clients. Scripting UAs with many distinct file IDs
  look more like automation than interactive work.
- Pivot raw `logs-anthropic.audit-*` for individual `anthropic.audit.claude_file_id` / filenames and whether
  `claude_file_viewed` or chat activity matches the upload volume.
- Correlate with artifact sharing, data exports, or compliance key creation from the same user.

### False positive analysis

- Document migrations and RAG indexing from one seat commonly exceed the threshold.

### Response and remediation

- On unauthorized staging: revoke sessions, review uploaded names/file IDs for sensitive content, and tighten project
  sharing or upload policy for the actor.
"""
references = ["https://platform.claude.com/docs/en/api/compliance/activities/list"]
risk_score = 47
rule_id = "58295a79-5dde-4fc9-a967-2dbe39fb8bc7"
severity = "medium"
tags = [
    "Domain: GenAI",
    "Platform: Anthropic",
    "Data Source: Anthropic Audit Logs",
    "Use Case: Threat Detection",
    "Use Case: UEBA",
    "Resources: Investigation Guide",
    "Rule Type: ES|QL",
    "Tactic: Exfiltration",
    "Mitre Atlas: AML.T0025",
]
timestamp_override = "event.ingested"
type = "esql"

query = '''
from logs-anthropic.audit-*
| where
    data_stream.dataset == "anthropic.audit" and
    event.action == "claude_file_uploaded" and
    event.outcome == "success" and
    user.email is not null and
    source.ip is not null
| stats
    Esql.event_count = count(*),
    Esql.event_id_values = values(event.id),
    Esql.file_name_values = values(file.name),
    Esql.anthropic_audit_claude_file_id_values = values(anthropic.audit.claude_file_id),
    Esql.anthropic_audit_claude_chat_id_values = values(anthropic.audit.claude_chat_id),
    Esql.anthropic_audit_claude_project_id_values = values(anthropic.audit.claude_project_id),
    Esql.user_agent_original_values = values(user_agent.original),
    Esql.anthropic_audit_actor_type_values = values(anthropic.audit.actor.type),
    Esql.timestamp_first_seen = min(@timestamp),
    Esql.timestamp_last_seen = max(@timestamp)
  by user.email, source.ip
| where Esql.event_count >= 100
| keep user.email, source.ip, Esql.*
'''


[[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_mappings]]
framework = "MITRE ATLAS"
version = "2026.08"
[[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.alert_suppression]
group_by = ["user.email", "source.ip"]
missing_fields_strategy = "suppress"

[rule.investigation_fields]
field_names = [
    "user.email",
    "source.ip",
    "Esql.event_count",
    "Esql.event_id_values",
    "Esql.file_name_values",
    "Esql.anthropic_audit_claude_file_id_values",
    "Esql.anthropic_audit_claude_chat_id_values",
    "Esql.anthropic_audit_claude_project_id_values",
    "Esql.user_agent_original_values",
    "Esql.anthropic_audit_actor_type_values",
    "Esql.timestamp_first_seen",
    "Esql.timestamp_last_seen",
]

[rule.alert_suppression.duration]
unit = "h"
value = 24

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.