Compromised accounts

When an attacker signs in as one of your people, everything they do looks like that person's work. Above investigates the account across identity, SaaS and endpoint and answers the first question: is this the employee, or someone using their access?

An office building with its front door shut. A ladder stands against the side wall, up to one open violet window.

INC-026A finance analyst's account

Confirmed threat

An intruder takes over a finance analyst's mailbox

  1. OktaNew device passes MFA
  2. Microsoft Entra IDNew app granted mailbox access
  3. OutlookNew bank details proposed

Response Reset sessions and MFA

Compromised accounts, by the numbers

  • 20%

    of insider incidents involve stolen credentials. Ponemon Institute, 2026 Cost of Insider Risks.

  • $842,462

    per credential-theft incident, the costliest insider type. Ponemon Institute, 2026 Cost of Insider Risks.

  • 13%

    of breaches begin with credential abuse. Verizon 2026 Data Breach Investigations Report.

Why compromised accounts are hard to call

“The sign-in was valid”

MFA was satisfied, the access was allowed and nothing broke a rule. On paper the attacker is the employee.

“The alert doesn't know the person”

An impossible-travel alert means little until you know whether the person is traveling, just changed roles or has never worked at that hour.

“We nearly blamed the employee”

Without context, the first suspect is the person whose name is on the account. Getting that wrong costs trust and slows the real response.

Not really Megan

Not really Megan

Above investigative agents

  • Timeline
  • Contextual analysis
  • Reasoning
  • Recommended actions

Ready for security, HR, and legal to act on together.

What the investigation hands you

Ready for security, HR, and legal to act on together.

Not really MeganINC-026 · A finance analyst's account

Confirmed threat

An intruder takes over a finance analyst's mailbox

Every step used Megan's valid sign-in, so nothing broke a rule.

The evidence

  • Person
  • Okta
  • Outlook
  • Microsoft Entra ID

A new device abroad after MFA fatigue, the forwarding rule, the app grant and the bank-details reply, against Megan's own MacBook and hours.

The response

  1. Reset sessions and MFA
  2. Remove the rule and the app
  3. Warn the vendor
  4. Clear Megan

A human confirms

A rep sends himself gift cards as a partner

At a Nasdaq-listed company, the partner portal recorded a partner issuing a gift card. Above showed it was a sales rep, and that the card went to him.

A Nasdaq-listed companyCustomer story

Talk to a human, 30 minutes

A solutions engineer walks you through a real case.

How Above investigates a compromised account

  1. Hours, devices, places and data, against the person's own history.

  2. MFA, devices and grants from your identity provider, in one timeline.

  3. The employee or someone else, and what to revoke and reset.

  • Okta
  • Microsoft Entra
  • Google Workspace
  • Ping
  • Microsoft Outlook
  • Google Drive
  • CrowdStrike
  • Microsoft Defender

Evidence this investigation reads

The signals Above reads for compromised accounts

Read together, they show whether the person behind the account changed.

Show the signals from
  • MFA fatigue and resets

    Okta, Microsoft Entra

    Repeated prompts, odd-hour approvals, reset methods.

  • New devices and locations

    Okta, Ping

    Never seen for this person, right before unusual activity.

  • Sessions in two places at once

    Okta, Google Workspace

    The owner working while another session runs elsewhere.

  • New OAuth grants after sign-in

    Microsoft Entra, Google Workspace

    Mail or file access granted minutes after a risky sign-in.

  • Inbox and forwarding rules

    Gmail, Microsoft Outlook

    Rules that hide replies or forward mail outside.

  • Downloads that break the pattern

    Google Drive

    Far beyond the person's own history.

  • Endpoint detections on the person's device

    CrowdStrike, Microsoft Defender

    From your EDR, in the same case.

  • Access to code and systems outside the role

    GitHub

    Repositories and admin tools the account never touched.

  • A risky install before the change

    CrowdStrike

    New software or an extension just before behavior shifts.

Clear the employee fast

In a compromised-account case the employee is usually the victim. Above's job is to show that quickly, so the response goes after the attacker and the person isn't treated as a suspect.

  • The case separates the owner's sessions from the others, with the evidence for each.
  • The person stays pseudonymized until a reviewer is authorized to unmask, and every unmask is logged.
  • Cases go to Jira, ServiceNow or CrowdStrike Falcon Next-Gen SIEM, so the SOC acts on the same timeline.

Related use cases

Shadow IT and unapproved software

Software, extensions and SaaS accounts IT never approved, and which ones carry risk.

  • Malicious insider and IP theft

    Legitimate access used to steal, leak or sabotage. Above separates intent from the job and builds the record.

  • AI agent insider threat

    What AI agents did with the access they were given, who they acted for, and whether it matched the ask.

Questions about compromised accounts

What is a compromised account?

A compromised account is a legitimate user account controlled by someone else, through phishing, a stolen session, MFA fatigue or malware. The access is real; the person using it isn't. The Ponemon Institute counts credential theft as one of three insider incident types, beside negligent and malicious insiders.

How do you tell a compromised account from a malicious insider?

You tell a compromised account from a malicious insider by comparing the activity with the person's own pattern. A new device, place or hours, behavior they have never shown, and the real owner working normally at the same time point to compromise. The same actions from their own laptop, during their day, point to the insider.

Isn't this the SOC's job?

The SOC owns the response. Above gives it the insider context it usually lacks: what the person normally does, which access they hold and whether the activity fits their job. The SOC decides faster, and the employee isn't wrongly blamed.

Does Above replace our identity threat detection?

Above works beside identity threat detection. It reads Okta, Entra, Google Workspace and Ping signals next to SaaS and endpoint evidence and investigates them as one case, so an identity alert arrives with what the person normally does, what the session touched and whether it fits their job.

Know if it's your employee, or an attacker

See a compromised-account investigation from first prompt to verdict.