Skip to main content

Ransomware response

How an AI SOC supports ransomware response: confirm encryption on the endpoint, scope hosts and accounts, page responders, and stage containment for approval.

Use case
Updated

The short answer

Ransomware response has to confirm encryption, find every affected host and account, and contain them before more data is lost. Cotool agents gather that evidence from EDR and identity tools and page responders, while containment waits for a person to approve it.[1][2][7]

The problem

Ransomware typically encrypts common user files such as Office documents, PDFs, images, and source code, often renaming or tagging them with file markers.[7]

Encryption often comes last: adversaries may first change file permissions or shut down systems to get at the files.[7]

How an AI SOC handles it

  1. Confirm on the endpoint

    From the EDR alert, the agent reads the SentinelOne threat and its timeline, or the CrowdStrike alert's behaviors, to confirm mass file changes.[2][3][8]

  2. Scope affected hosts

    The agent hunts for the same process or hash across endpoints with SentinelOne PowerQuery or CrowdStrike Next-Gen SIEM.[3][9]

  3. Check the accounts

    The user and service accounts on each host are checked in Okta or Entra ID for unusual sign-ins.[3][4]

  4. Page responders

    The agent escalates and follows your escalation guidance, for example paging on-call through PagerDuty for a SEV-1.[3][5]

  5. Stage containment

    The agent prepares network containment for each affected host and session revocation for exposed accounts, each waiting for approval.[1][2][10][11]

Example

Count SentinelOne events by typetext
dataSource.name='SentinelOne' dataSource.category='security' event.type=*
| group ct=count() by event.type
| sort -ct
| limit 50

A PowerQuery that groups security events by type over the query window, a starting point for finding hosts with unusual file activity. SentinelOne runs it through its asynchronous query API.[9]

What people approve

  • Isolating hosts and revoking sessions are state-changing actions that can require approval in the app or in Slack.[1]
  • Response agents cannot close an alert as True Positive. When an agent confirms malicious activity, it escalates the alert and a person closes it.[6]

How Cotool handles it

  • Cotool agents can hand an investigation to parallel investigator subagents, such as one for the endpoint and one for identity, with a reviewer confirming the verdict.[1][2]

Integrations used

Sources

  1. AI SOC (Cotool) · accessed November 16, 2026
  2. Respond to every signal (Cotool Respond) · accessed November 16, 2026
  3. Changelog (Cotool docs) · accessed November 16, 2026
  4. Microsoft Graph integration (Cotool docs) · accessed November 16, 2026
  5. Context documents (Cotool docs) · accessed November 16, 2026
  6. Alerts (Cotool docs) · accessed November 16, 2026
  7. Data Encrypted for Impact, T1486 (MITRE ATT&CK) · accessed November 16, 2026
  8. Threats (SentinelOne management API reference) · accessed November 16, 2026
  9. LRQ API reference (SentinelOne ai-siem) · accessed November 16, 2026
  10. Hosts (FalconPy) · accessed November 16, 2026
  11. Agent Actions (SentinelOne management API reference) · accessed November 16, 2026

Related use cases

All use cases

Related reading

Attackers are scaling with tokens

Cotool helps defenders operate at machine speed. See how security teams are scaling Detection & Response.

Book a demo