1. Resources
  2. Detection rules
  3. PingOne Protect Risk Policy or Predictor Changed

PingOne Protect Risk Policy or Predictor Changed detection rules

An attacker with PingOne administrator rights edits, reorders or deletes a risk policy or a risk predictor of PingOne Protect, so that sign-ons from their address or device are no longer rated as risky and pass without a challenge.

Platform
Ping Identity
Severity
Medium
Log source
PingOne audit activities
Last updated
Rule formats
Sigmaand queries for: SplunkMicrosoft Sentinel
What an attacker needs
A PingOne role that can edit PingOne Protect configuration, and sign-on policies or flows that rely on the risk level.
Impact
Sign-ons that PingOne Protect would have rated as high risk are rated low and pass with fewer factors in every flow that uses the policy.
Response
Restore the risk policy sets and predictors from the reviewed copy and review the sign-ons evaluated while the weaker configuration applied.
Mitigation
Limit the roles that can edit PingOne Protect configuration, and require MFA for administrators whatever the risk level.
Action types
RISK_POLICY_SET.CREATEDRISK_POLICY_SET.UPDATEDRISK_POLICY_SET.DELETEDRISK_POLICY_SET.ORDER_UPDATEDRISK_PREDICTOR.UPDATEDRISK_PREDICTOR.DELETEDRISK_SETTINGS.UPDATED

How the attack works

PingOne Protect rates a sign-on before the flow decides what to ask for. A risk evaluation runs the risk predictors of the environment, such as anonymous network detection, IP reputation, IP velocity, new device, bot detection and adversary-in-the-middle detection, and a risk policy set combines their outcomes into a result with the level HIGH, MEDIUM or LOW. Sign-on policies and flows use that level to step up to MFA or to deny access. The rating is configuration, and each part of it can be bent. A predictor can carry a list of IP ranges it ignores. A policy can hold IP range conditions and override policies, which Ping documents as the ones with the highest priority. Targeted policies are processed in order, and the first one whose conditions match the transaction type, user group or application is used. One policy set is the environment's default.

An attacker who can edit these adds their own address range to the ignored list, puts an override in front, moves a permissive targeted policy to the top, or deletes the predictor that would have caught them. Nothing about any user changes, and every later sign-on that matches is rated low. PingOne records the changes as RISK_POLICY_SET, RISK_PREDICTOR and RISK_SETTINGS audit activities. The page follows Ping's documentation of PingOne Protect: no public incident report describes these changes for PingOne.

Log source: the rules read PingOne audit activities. Include the Risk Policy, Risk Predictor and Risk Settings event types in the PingOne webhook that feeds the SIEM, or poll the Audit Activities API, for every environment that uses PingOne Protect.

What an attacker needs

  • A PingOne role that can edit PingOne Protect configuration in the environment, built-in or custom.
  • Sign-on policies or flows that rely on the risk level to decide when MFA is required or access is denied.

Impact

Sign-ons that PingOne Protect would have rated as high risk, for example from an anonymizing network, a new device or a reverse proxy between user and service, are rated low and pass with fewer factors. A stolen password or session becomes usable where the risk check was the control that stopped it, in every application whose flow uses the changed policy.

Detection rules

PingOne Risk Policy Created, Updated, Deleted or Reordered

Detects a change to the risk policy sets of PingOne Protect (audit activities RISK_POLICY_SET.CREATED, RISK_POLICY_SET.UPDATED, RISK_POLICY_SET.DELETED and RISK_POLICY_SET.ORDER_UPDATED, described by Ping as Risk Policy Created, Updated, Deleted and Risk Policy Order Updated). A risk policy set turns the outcomes of the risk predictors into the level that sign-on policies and flows act on. Ping documents that a policy can hold IP range conditions and override policies with the highest priority, that one policy set is the environment's default, and that targeted policies are processed in a set order in which the first matching one is used. An added exemption, a new targeted policy or a new order can rate an attacker's sign-ons as low risk. The activity does not say what changed or whether the result is more lenient: compare the policy set and the order with your reviewed copy.

LevelMediumStatusExperimentalLog sourcePingOne audit activities
title: PingOne Risk Policy Created, Updated, Deleted or Reordered
id: 4e4dd108-7036-4f3e-9d40-9d4102bfa406
status: experimental
description: >-
  Detects a change to the risk policy sets of PingOne Protect (audit
  activities RISK_POLICY_SET.CREATED, RISK_POLICY_SET.UPDATED,
  RISK_POLICY_SET.DELETED and RISK_POLICY_SET.ORDER_UPDATED, described by Ping
  as Risk Policy Created, Updated, Deleted and Risk Policy Order Updated). A
  risk policy set turns the outcomes of the risk predictors into the level
  that sign-on policies and flows act on. Ping documents that a policy can
  hold IP range conditions and override policies with the highest priority,
  that one policy set is the environment's default, and that targeted
  policies are processed in a set order in which the first matching one is
  used. An added exemption, a new targeted policy or a new order can rate an
  attacker's sign-ons as low risk. The activity does not say what changed or
  whether the result is more lenient: compare the policy set and the order
  with your reviewed copy.
references:
  - https://attack.mitre.org/techniques/T1556/009/
  - https://developer.pingidentity.com/pingone-api/platform/reference/audit-reporting-events.html
  - https://developer.pingidentity.com/pingone-api/protect/risk-policies/index.html
author: Forestall
date: 2026-11-18
tags:
  - attack.defense-impairment
  - attack.persistence
  - attack.credential-access
  - attack.t1556.009
logsource:
  product: pingone
  service: audit
detection:
  selection:
    action.type:
      - RISK_POLICY_SET.CREATED
      - RISK_POLICY_SET.UPDATED
      - RISK_POLICY_SET.DELETED
      - RISK_POLICY_SET.ORDER_UPDATED
  condition: selection
falsepositives:
  - Tuning of risk policies by the team that owns authentication, for example new scores after a review of false positives, or a new targeted policy for one application. Confirm with the change record.
level: medium

False positives

  • Tuning of risk policies by the team that owns authentication, for example new scores after a review of false positives, or a new targeted policy for one application. Confirm with the change record.

PingOne Risk Predictor Updated or Deleted

Detects a change to a risk predictor of PingOne Protect (audit activities RISK_PREDICTOR.UPDATED and RISK_PREDICTOR.DELETED, described by Ping as Risk Predictor Updated and Deleted). Predictors are the checks a risk evaluation runs, such as anonymous network detection, IP reputation, IP velocity, new device, bot detection and adversary-in-the-middle detection. Ping documents that several of them accept a list of IP ranges the predictor ignores, and that a predictor has a default result level. An attacker who adds their own range to that list, or deletes the predictor, removes the check for their sign-ons while everything else looks unchanged. The activity does not say what changed: read the predictor and compare it with your reviewed copy.

LevelMediumStatusExperimentalLog sourcePingOne audit activities
title: PingOne Risk Predictor Updated or Deleted
id: bc56c2df-fa21-45b4-85be-fc7c160c7454
status: experimental
description: >-
  Detects a change to a risk predictor of PingOne Protect (audit activities
  RISK_PREDICTOR.UPDATED and RISK_PREDICTOR.DELETED, described by Ping as Risk
  Predictor Updated and Deleted). Predictors are the checks a risk evaluation
  runs, such as anonymous network detection, IP reputation, IP velocity, new
  device, bot detection and adversary-in-the-middle detection. Ping documents
  that several of them accept a list of IP ranges the predictor ignores, and
  that a predictor has a default result level. An attacker who adds their own
  range to that list, or deletes the predictor, removes the check for their
  sign-ons while everything else looks unchanged. The activity does not say
  what changed: read the predictor and compare it with your reviewed copy.
references:
  - https://attack.mitre.org/techniques/T1556/009/
  - https://developer.pingidentity.com/pingone-api/platform/reference/audit-reporting-events.html
  - https://developer.pingidentity.com/pingone-api/protect/risk-predictors/index.html
author: Forestall
date: 2026-11-18
tags:
  - attack.defense-impairment
  - attack.persistence
  - attack.credential-access
  - attack.t1556.009
logsource:
  product: pingone
  service: audit
detection:
  selection:
    action.type:
      - RISK_PREDICTOR.UPDATED
      - RISK_PREDICTOR.DELETED
  condition: selection
falsepositives:
  - Planned changes, such as adding the address range of a new office or VPN to the ignored networks, or removing a custom predictor that is no longer used. Confirm with the change record and check that every added range belongs to the organization.
level: medium

False positives

  • Planned changes, such as adding the address range of a new office or VPN to the ignored networks, or removing a custom predictor that is no longer used. Confirm with the change record and check that every added range belongs to the organization.

PingOne Protect Settings Updated

Detects a change to the Protect settings of a PingOne environment (audit activity RISK_SETTINGS.UPDATED, described by Ping as Risk Settings Updated). Ping documents these settings as the maximum data retention periods for the risk data of three predictors: New Device, User Location Anomaly and User-based Risk Behavior. For the New Device predictor the period can be 1 to 180 days, with 180 as the default. A new period applies only to data collected after the change. A much shorter period makes these predictors forget known devices and usual locations sooner, which weakens what they can compare a sign-on with. The effect is slow, so the rule is for review and as context for the other rules of this page.

LevelLowStatusExperimentalLog sourcePingOne audit activities
title: PingOne Protect Settings Updated
id: 00b337be-0ce0-4893-aa66-764a2bb41e2f
status: experimental
description: >-
  Detects a change to the Protect settings of a PingOne environment (audit
  activity RISK_SETTINGS.UPDATED, described by Ping as Risk Settings Updated).
  Ping documents these settings as the maximum data retention periods for the
  risk data of three predictors: New Device, User Location Anomaly and
  User-based Risk Behavior. For the New Device predictor the period can be 1
  to 180 days, with 180 as the default. A new period applies only to data
  collected after the change. A much shorter period makes these predictors forget known devices
  and usual locations sooner, which weakens what they can compare a sign-on
  with. The effect is slow, so the rule is for review and as context for the
  other rules of this page.
references:
  - https://attack.mitre.org/techniques/T1556/009/
  - https://developer.pingidentity.com/pingone-api/platform/reference/audit-reporting-events.html
  - https://developer.pingidentity.com/pingone-api/protect/protect-settings/index.html
author: Forestall
date: 2026-11-18
tags:
  - attack.defense-impairment
  - attack.persistence
  - attack.credential-access
  - attack.t1556.009
logsource:
  product: pingone
  service: audit
detection:
  selection:
    action.type: RISK_SETTINGS.UPDATED
  condition: selection
falsepositives:
  - A change of retention periods for privacy or compliance reasons. The settings change rarely, so confirm each activity with the change record.
level: low

False positives

  • A change of retention periods for privacy or compliance reasons. The settings change rarely, so confirm each activity with the change record.

Detection logic

Patterns that need more than one event, a baseline or knowledge of your own environment. Each is a hypothesis to build and tune in your SIEM, not a ready rule.

Share of high risk evaluations drops after a risk policy or predictor change

Detection hypothesisNeeds a baselineCompares with an earlier value

A change that exempts an attacker also changes the numbers. If the share of risk evaluations with the level HIGH or MEDIUM falls right after a policy or predictor activity while the volume of sign-ons stays the same, the change made the rating more lenient, whatever its stated purpose was.

  1. Note each RISK_POLICY_SET or RISK_PREDICTOR activity that updates, deletes or reorders, with its time.
  2. Count RISK_EVALUATION.CREATED activities per day and per _embedded.riskResult.level, which Ping's event list names as an attribute of the activity. The Risk Evaluations API documents the levels HIGH, MEDIUM and LOW.
  3. Compare the share of each level in the 7 days before the change with the days after it, for the environment and for the applications that use the changed policy set.

EventsRISK_EVALUATION.CREATEDRISK_POLICY_SET.UPDATEDRISK_PREDICTOR.UPDATED

LimitsPing's event list gives the level attribute no description, so check it against one evaluation you can also read through the API. The share moves with holidays, new applications and real attacks as well, so a drop is a reason to read the configuration, not proof. A narrow exemption for one address does not move the totals.

Triage

  • Identify the actor (actors.user.name or actors.client.name) and the policy set or predictor named in the activity's resources, and check for an approved change.
  • Read the policy set or predictor through the API or the console and compare it with the reviewed copy. Look for new ignored IP ranges, new IP range conditions, a new override policy, a changed default result, a changed order of targeted policies and a predictor that is missing.
  • Check which applications and flows use the policy set, and whether it is the environment's default.
  • Compare the addresses that were added with the source address of the change (source.ipAddress) and with recent sign-ons.
  • Review sign-ons since the change that would have matched the removed or weakened check, and what else the actor changed in the same session.

Response

Once the attack is confirmed:

  1. Restore the policy set, the order of targeted policies and the predictors from the reviewed copy, and recreate a predictor that was deleted.
  2. Suspend the administrator or worker application that made the change and remove the role that allows it.
  3. Review the sign-ons that were evaluated while the weaker configuration was in place, and end the sessions of accounts that signed on from the ranges or conditions the change exempted.
  4. Reset the credentials of accounts whose owners cannot confirm those sign-ons.
  5. Compare every risk policy set and predictor of the environment with the reviewed copy, and check the sign-on policies and flows that consume the risk level.

Mitigation

  • Limit the roles that can edit PingOne Protect configuration, and require a change record for every edit.
  • Keep a reviewed copy of every risk policy set, the order of targeted policies, and every predictor with its ignored IP ranges, and compare it after every change event.
  • Do not let risk be the only gate. Require MFA for administrators and for sensitive applications at every sign-on, whatever the risk level.
  • Keep the lists of ignored networks short and tied to addresses the organization owns.
  • Alert on these event types in every environment that uses PingOne Protect.

Frequently asked questions

Which changes to a PingOne Protect configuration lower the risk rating of a sign-on?

Ping documents several. A predictor such as IP reputation or anonymous network detection can hold a list of IP ranges that it ignores. A policy can contain IP range conditions and override policies, which have the highest priority. Targeted policies are processed in a set order and the first matching one is used, so a reorder changes which policy applies. A deleted predictor no longer contributes at all.

Does the audit activity show what was changed in the policy or predictor?

No. Ping's event list gives the risk policy and predictor activities the common attributes only, such as the actor, the source address and the resource. Which condition, range or score changed is not in a documented attribute. Keep a reviewed copy of the configuration and compare it after each alert.

Why is a change to the Protect settings rated lower than the other two rules?

Ping documents the Protect settings as the data retention periods of three predictors: New Device, User Location Anomaly and User-based Risk Behavior. A new period applies only to data collected after the change. A shorter period makes the predictors forget sooner, which weakens them slowly, while a changed policy or predictor takes effect at the next sign-on.

Has tampering with PingOne Protect been reported in an attack?

No public incident report describes it. The page is derived from Ping's documentation of risk policies, predictors and settings, and it mirrors what MITRE ATT&CK describes for conditional access policies at other identity providers, where adding trusted IP ranges to a policy is a way to make later sign-ons pass.

References

We respect your privacy

We use cookies to keep this site secure and working properly. With your permission, we also use optional cookies to understand usage and improve the experience. Cookie Policy

You can change your choice at any time.