← Back to Explore
elasticmediumTTP
Anthropic Multiple Authentication Failures
Detects at least five failed Anthropic authentication events for the same user email within one hour. Failures are matched by authentication category and failure outcome (for example magic-link or SSO login failures). That pattern fits repeated guessing, stale magic link abuse, or automated login attempts against one account.
Detection Query
from logs-anthropic.audit-*
| where
data_stream.dataset == "anthropic.audit" and
mv_contains(event.category, "authentication") and
event.outcome == "failure" and
user.email is not null
| stats
Esql.event_count = count(*),
Esql.event_id_values = values(event.id),
Esql.event_action_values = values(event.action),
Esql.source_ip_values = values(source.ip),
Esql.source_ip_distinct_count = count_distinct(source.ip),
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
| where Esql.event_count >= 5
| keep user.email, Esql.*
Author
Elastic
Created
2026/09/12
Data Sources
Anthropic Audit Logs
Tags
Domain: GenAIDomain: IdentityPlatform: AnthropicData Source: Anthropic Audit LogsUse Case: Identity and Access AuditUse Case: Threat DetectionUse Case: UEBAResources: Investigation GuideRule Type: ES|QLTactic: Credential Access
Raw Content
[metadata]
creation_date = "2026/09/12"
integration = ["anthropic"]
maturity = "production"
updated_date = "2026/09/25"
[rule]
author = ["Elastic"]
description = """
Detects at least five failed Anthropic authentication events for the same user email within one hour. Failures are
matched by authentication category and failure outcome (for example magic-link or SSO login failures). That pattern fits
repeated guessing, stale magic link abuse, or automated login attempts against one account.
"""
false_positives = [
"""
A user repeatedly clicking an expired or invalid magic link from the same browser session can produce several
failures before requesting a new link or signing in successfully.
""",
"""
IdP or SSO cutover testing against a pilot account can generate a short burst of failed attempts for one email
during maintenance windows.
""",
]
from = "now-60m"
interval = "15m"
language = "esql"
license = "Elastic License v2"
name = "Anthropic Multiple Authentication Failures"
note = """## Triage and analysis
### Investigating Anthropic Multiple Authentication Failures
One mailbox produced a burst of authentication failures (magic-link, SSO, or other auth failures) within an hour.
Use IP diversity on the alert to separate focused retries from distributed automation.
Escalate when many distinct IPs hit one email, failures precede a successful login from an unfamiliar IP, or SSO /
magic-link second-factor weakening sits nearby. Close as FP for expired magic-link retries from one IP/UA or IdP
pilot testing with a ticket.
#### Possible investigation steps
- Read `Esql.event_count`, `Esql.event_action_values`, `Esql.source_ip_values`, and
`Esql.source_ip_distinct_count`.
- Low IP diversity → one device/egress (typos, expired links, focused guessing). High diversity → distributed
automation or proxy rotation — higher priority.
- Inspect UA / actor-type values for automation. Search for successful `magic_link_login_succeeded` /
`sso_login_succeeded` for the same email or IPs in the window.
- Correlate with **Anthropic SSO Disabled or Connection Removed** or magic-link second-factor disablement when
org-scoped admin events are present.
### False positive analysis
- Expired magic-link click storms from one IP and named IdP cutover tests are common FPs.
### Response and remediation
- On suspected stuffing/takeover: invalidate sessions, reset credentials/MFA, and review successful logins from new
IPs. If failures precede a successful unfamiliar login, expand to data access and admin changes.
"""
references = ["https://platform.claude.com/docs/en/api/compliance/activities/list"]
risk_score = 47
rule_id = "25647990-cc37-430c-9a4e-2793bb5445f5"
severity = "medium"
tags = [
"Domain: GenAI",
"Domain: Identity",
"Platform: Anthropic",
"Data Source: Anthropic Audit Logs",
"Use Case: Identity and Access Audit",
"Use Case: Threat Detection",
"Use Case: UEBA",
"Resources: Investigation Guide",
"Rule Type: ES|QL",
"Tactic: Credential Access",
]
timestamp_override = "event.ingested"
type = "esql"
query = '''
from logs-anthropic.audit-*
| where
data_stream.dataset == "anthropic.audit" and
mv_contains(event.category, "authentication") and
event.outcome == "failure" and
user.email is not null
| stats
Esql.event_count = count(*),
Esql.event_id_values = values(event.id),
Esql.event_action_values = values(event.action),
Esql.source_ip_values = values(source.ip),
Esql.source_ip_distinct_count = count_distinct(source.ip),
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
| where Esql.event_count >= 5
| keep user.email, Esql.*
'''
[[rule.threat]]
framework = "MITRE ATT&CK"
[[rule.threat.technique]]
id = "T1110"
name = "Brute Force"
reference = "https://attack.mitre.org/techniques/T1110/"
[rule.threat.tactic]
id = "TA0006"
name = "Credential Access"
reference = "https://attack.mitre.org/tactics/TA0006/"
[[rule.threat_mappings]]
framework = "MITRE ATLAS"
version = "2026.08"
[[rule.threat_mappings.threat]]
framework = "MITRE ATLAS"
[rule.threat_mappings.threat.tactic]
id = "AML.TA0013"
name = "Credential Access"
reference = "https://atlas.mitre.org/tactics/AML.TA0013/"
[rule.alert_suppression]
group_by = ["user.email"]
missing_fields_strategy = "suppress"
[rule.investigation_fields]
field_names = [
"user.email",
"Esql.event_count",
"Esql.event_id_values",
"Esql.event_action_values",
"Esql.source_ip_values",
"Esql.source_ip_distinct_count",
"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 = 1