Security Command Center helps you detect and investigate threats to your AI workloads and
AI agents. You can review findings from detectors that monitor general
AI-related threats and from detectors designed for AI agents that are deployed
to
Agent Runtime Runtime.

## General AI threats

The following log-based detections are available with
[Event Threat Detection](https://docs.cloud.google.com/security-command-center/docs/concepts-event-threat-detection-overview.md.txt):

<br />

- [`Impact: Gemini Discovery Engine API Document Import by External User`](https://docs.cloud.google.com/security-command-center/docs/findings/threats/gemini-discovery-engine-api-document-import-by-external-user.md.txt)
- [`Impact: Gemini Discovery Engine API Document Import from External Resource`](https://docs.cloud.google.com/security-command-center/docs/findings/threats/gemini-discovery-engine-api-document-import-from-external-resource.md.txt)
- [`Impact: Gemini Discovery Engine API Document Import VPC Perimeter Violation`](https://docs.cloud.google.com/security-command-center/docs/findings/threats/gemini-discovery-engine-api-document-import-vpc-perimeter-violation.md.txt)
- [`Initial Access: Dormant Service Account Activity in AI Service`](https://docs.cloud.google.com/security-command-center/docs/findings/threats/ai-dormant-service-account-used-in-action.md.txt)
- [`Persistence: External Access Granted to Gemini Discovery Engine API Apps`](https://docs.cloud.google.com/security-command-center/docs/findings/threats/external-access-granted-to-gemini-discovery-engine-api-apps.md.txt)
- [`Persistence: New AI API Method`](https://docs.cloud.google.com/security-command-center/docs/findings/threats/ai-anomalous-behavior-new-api-method.md.txt)
- [`Persistence: New Geography for AI Service`](https://docs.cloud.google.com/security-command-center/docs/findings/threats/ai-iam-anomalous-behavior-ip-geolocation.md.txt)
- [`Privilege Escalation: Anomalous Impersonation of Service Account for AI Admin Activity`](https://docs.cloud.google.com/security-command-center/docs/findings/threats/ai-anomalous-sa-delegation-impersonation-of-sa-admin-activity.md.txt)
- [`Privilege Escalation: Anomalous Multistep Service Account Delegation for AI Admin Activity`](https://docs.cloud.google.com/security-command-center/docs/findings/threats/ai-anomalous-sa-delegation-multistep-admin-activity.md.txt)
- [`Privilege Escalation: Anomalous Multistep Service Account Delegation for AI Data Access`](https://docs.cloud.google.com/security-command-center/docs/findings/threats/ai-anomalous-sa-delegation-multistep-data-access.md.txt)
- [`Privilege Escalation: Anomalous Service Account Impersonator for AI Admin Activity`](https://docs.cloud.google.com/security-command-center/docs/findings/threats/ai-anomalous-sa-delegation-impersonator-admin-activity.md.txt)
- [`Privilege Escalation: Anomalous Service Account Impersonator for AI Data Access`](https://docs.cloud.google.com/security-command-center/docs/findings/threats/ai-anomalous-sa-delegation-impersonator-data-access.md.txt)

<br />

## Threats to agents deployed to Agent Runtime Runtime


> [!WARNING]
>
> **Preview**
>
>
> This feature is
>
> subject to the "Pre-GA Offerings Terms" in the General Service Terms section of the
> [Service Specific
> Terms](https://docs.cloud.google.com/terms/service-terms#1).
>
> Pre-GA features are available "as is" and might have limited support.
>
> For more information, see the
> [launch stage descriptions](https://cloud.google.com/products/#product-launch-stages).

<br />

Security Command Center performs [runtime](https://docs.cloud.google.com/security-command-center/docs/ai-threats.md.txt#runtime) and [control plane](https://docs.cloud.google.com/security-command-center/docs/ai-threats.md.txt#control-plane)
monitoring of AI agents deployed to
[Agent Runtime](https://docs.cloud.google.com/agent-builder/agent-engine/overview) Runtime.

### Runtime finding types

The following runtime detections are available with
[Agent Platform Threat Detection](https://docs.cloud.google.com/security-command-center/docs/agent-engine-threat-detection-overview):

<br />

- [`Command and Control: Steganography Tool Detected`](https://docs.cloud.google.com/security-command-center/docs/findings/threats/agent-platform-steganography-tool-detected.md.txt)
- [`Credential Access: Find Google Cloud Credentials`](https://docs.cloud.google.com/security-command-center/docs/findings/threats/agent-platform-find-gcp-credentials.md.txt)
- [`Credential Access: GPG Key Reconnaissance`](https://docs.cloud.google.com/security-command-center/docs/findings/threats/agent-platform-gpg-key-reconnaissance.md.txt)
- [`Credential Access: Search Private Keys or Passwords`](https://docs.cloud.google.com/security-command-center/docs/findings/threats/agent-platform-search-private-keys-or-passwords.md.txt)
- [`Defense Evasion: Base64 ELF File Command Line`](https://docs.cloud.google.com/security-command-center/docs/findings/threats/agent-platform-base64-elf-file-cmdline.md.txt)
- [`Defense Evasion: Base64 Encoded Python Script Executed`](https://docs.cloud.google.com/security-command-center/docs/findings/threats/agent-platform-base64-encoded-python-script-executed.md.txt)
- [`Defense Evasion: Base64 Encoded Shell Script Executed`](https://docs.cloud.google.com/security-command-center/docs/findings/threats/agent-platform-base64-encoded-shell-script-executed.md.txt)
- [`Defense Evasion: Launch Code Compiler Tool In Container`](https://docs.cloud.google.com/security-command-center/docs/findings/threats/agent-platform-launch-code-compiler-tool-in-container.md.txt)
- [`Execution: Added Malicious Binary Executed`](https://docs.cloud.google.com/security-command-center/docs/findings/threats/agent-platform-added-malicious-binary-executed.md.txt)
- [`Execution: Added Malicious Library Loaded`](https://docs.cloud.google.com/security-command-center/docs/findings/threats/agent-platform-added-malicious-library-loaded.md.txt)
- [`Execution: Added Malicious Skill Executed`](https://docs.cloud.google.com/security-command-center/docs/findings/threats/agent-platform-added-malicious-skill-executed.md.txt)
- [`Execution: Added Malicious Skill Loaded`](https://docs.cloud.google.com/security-command-center/docs/findings/threats/agent-platform-added-malicious-skill-loaded.md.txt)
- [`Execution: Built in Malicious Binary Executed`](https://docs.cloud.google.com/security-command-center/docs/findings/threats/agent-platform-built-in-malicious-binary-executed.md.txt)
- [`Execution: Container Escape`](https://docs.cloud.google.com/security-command-center/docs/findings/threats/agent-platform-container-escape.md.txt)
- [`Execution: Fileless Execution in /memfd:`](https://docs.cloud.google.com/security-command-center/docs/findings/threats/agent-platform-fileless-execution-detection-memfd.md.txt)
- [`Execution: Kubernetes Attack Tool Execution`](https://docs.cloud.google.com/security-command-center/docs/findings/threats/agent-platform-kubernetes-attack-tool-execution.md.txt)
- [`Execution: Local Reconnaissance Tool Execution`](https://docs.cloud.google.com/security-command-center/docs/findings/threats/agent-platform-local-reconnaissance-tool-execution.md.txt)
- [`Execution: Malicious Python Executed`](https://docs.cloud.google.com/security-command-center/docs/findings/threats/agent-platform-malicious-python-executed.md.txt)
- [`Execution: Modified Malicious Binary Executed`](https://docs.cloud.google.com/security-command-center/docs/findings/threats/agent-platform-modified-malicious-binary-executed.md.txt)
- [`Execution: Modified Malicious Library Loaded`](https://docs.cloud.google.com/security-command-center/docs/findings/threats/agent-platform-modified-malicious-library-loaded.md.txt)
- [`Execution: Modified Malicious Skill Executed`](https://docs.cloud.google.com/security-command-center/docs/findings/threats/agent-platform-modified-malicious-skill-executed.md.txt)
- [`Execution: Modified Malicious Skill Loaded`](https://docs.cloud.google.com/security-command-center/docs/findings/threats/agent-platform-modified-malicious-skill-loaded.md.txt)
- [`Execution: Netcat Remote Code Execution in Container`](https://docs.cloud.google.com/security-command-center/docs/findings/threats/agent-platform-netcat-remote-code-execution-in-container.md.txt)
- [`Execution: Possible Arbitrary Command Execution through CUPS (CVE-2024-47177)`](https://docs.cloud.google.com/security-command-center/docs/findings/threats/agent-platform-possible-arbitrary-command-execution-through-cups.md.txt)
- [`Execution: Possible Remote Command Execution Detected`](https://docs.cloud.google.com/security-command-center/docs/findings/threats/agent-platform-possible-remote-command-execution-detected.md.txt)
- [`Execution: Program Run with Disallowed HTTP Proxy Env`](https://docs.cloud.google.com/security-command-center/docs/findings/threats/agent-platform-program-run-with-disallowed-http-proxy-env.md.txt)
- [`Execution: Socat Reverse Shell Detected`](https://docs.cloud.google.com/security-command-center/docs/findings/threats/agent-platform-socat-reverse-shell-detected.md.txt)
- [`Execution: Suspicious OpenSSL Shared Object Loaded`](https://docs.cloud.google.com/security-command-center/docs/findings/threats/agent-platform-suspicious-openssl-shared-object-loaded.md.txt)
- [`Exfiltration: Launch Remote File Copy Tools in Container`](https://docs.cloud.google.com/security-command-center/docs/findings/threats/agent-platform-launch-remote-file-copy-tools-in-container.md.txt)
- [`Impact: Detect Malicious Cmdlines`](https://docs.cloud.google.com/security-command-center/docs/findings/threats/agent-platform-detect-malicious-cmdlines.md.txt)
- [`Impact: Remove Bulk Data from Disk`](https://docs.cloud.google.com/security-command-center/docs/findings/threats/agent-platform-remove-bulk-data-from-disk.md.txt)
- [`Impact: Suspicious crypto mining activity using the Stratum Protocol`](https://docs.cloud.google.com/security-command-center/docs/findings/threats/agent-platform-detect-crypto-miners-using-stratum-protocol.md.txt)
- [`Malicious Script Executed`](https://docs.cloud.google.com/security-command-center/docs/findings/threats/agent-platform-malicious-script-executed.md.txt)
- [`Malicious URL Observed`](https://docs.cloud.google.com/security-command-center/docs/findings/threats/agent-platform-malicious-url-observed.md.txt)
- [`Privilege Escalation: Attempt to Abuse Sudo For Privilege Escalation (CVE-2019-14287)`](https://docs.cloud.google.com/security-command-center/docs/findings/threats/agent-platform-abuse-sudo-for-privilege-escalation.md.txt)
- [`Privilege Escalation: Polkit Local Privilege Escalation Vulnerability (CVE-2021-4034)`](https://docs.cloud.google.com/security-command-center/docs/findings/threats/agent-platform-polkit-local-privilege-escalation-vulnerability.md.txt)
- [`Privilege Escalation: Sudo Potential Privilege Escalation (CVE-2021-3156)`](https://docs.cloud.google.com/security-command-center/docs/findings/threats/agent-platform-sudo-potential-privilege-escalation.md.txt)
- [`Reverse Shell`](https://docs.cloud.google.com/security-command-center/docs/findings/threats/agent-platform-reverse-shell.md.txt)
- [`Unexpected Child Shell`](https://docs.cloud.google.com/security-command-center/docs/findings/threats/agent-platform-unexpected-child-shell.md.txt)

<br />

### Control plane finding types

The following control plane detections are available with
[Event Threat Detection](https://docs.cloud.google.com/security-command-center/docs/concepts-event-threat-detection-overview.md.txt):

<br />

- [`Credential Access: AI Agent Anomalous Access to Metadata Service`](https://docs.cloud.google.com/security-command-center/docs/findings/threats/ai-agent-anomalous-access-to-metadata-service.md.txt)
- [`Defense Evasion: Folder Level TokenCreator Role Granted to AI Agent`](https://docs.cloud.google.com/security-command-center/docs/findings/threats/ai-agent-granted-token-creator-role-folder.md.txt)
- [`Defense Evasion: Organization Level TokenCreator Role Granted to AI Agent`](https://docs.cloud.google.com/security-command-center/docs/findings/threats/ai-agent-granted-token-creator-role-org.md.txt)
- [`Defense Evasion: Project Level TokenCreator Role Granted to AI Agent`](https://docs.cloud.google.com/security-command-center/docs/findings/threats/ai-agent-granted-token-creator-role-project.md.txt)
- [`Discovery: AI Agent Service Account Self-Investigation`](https://docs.cloud.google.com/security-command-center/docs/findings/threats/ai-agent-iam-anomalous-behavior-service-account-gets-own-iam-policy.md.txt)
- [`Discovery: AI Agent Unauthorized Service Account API Call`](https://docs.cloud.google.com/security-command-center/docs/findings/threats/ai-agent-unauthorized-service-account-api-call.md.txt)
- [`Discovery: Evidence of Port Scanning from AI Agent`](https://docs.cloud.google.com/security-command-center/docs/findings/threats/ai-agent-port-scanning-evidence.md.txt)
- [`Exfiltration: AI Agent Initiated BigQuery Data Exfiltration to External Table`](https://docs.cloud.google.com/security-command-center/docs/findings/threats/ai-agent-big-query-exfil.md.txt)
- [`Exfiltration: AI Agent Initiated BigQuery Data Extraction`](https://docs.cloud.google.com/security-command-center/docs/findings/threats/ai-agent-big-query-exfil-to-cloud-storage.md.txt)
- [`Exfiltration: AI Agent Initiated BigQuery VPC Perimeter Violation`](https://docs.cloud.google.com/security-command-center/docs/findings/threats/ai-agent-big-query-vpc-violation.md.txt)
- [`Exfiltration: AI Agent Initiated CloudSQL Exfiltration to External Bucket`](https://docs.cloud.google.com/security-command-center/docs/findings/threats/ai-agent-cloudsql-exfil-export-external.md.txt)
- [`Exfiltration: AI Agent Initiated CloudSQL Exfiltration to Public Bucket`](https://docs.cloud.google.com/security-command-center/docs/findings/threats/ai-agent-cloudsql-exfil-export-public.md.txt)
- [`Initial Access: AI Agent Identity Excessive Permission Denied Actions`](https://docs.cloud.google.com/security-command-center/docs/findings/threats/ai-agent-excessive-failed-attempt.md.txt)
- [`Persistence: Sensitive AI Permission Added to Custom Role`](https://docs.cloud.google.com/security-command-center/docs/findings/threats/sensitive-ai-permission-added-to-custom-role.md.txt)
- [`Persistence: Sensitive Role Granted by AI Agent`](https://docs.cloud.google.com/security-command-center/docs/findings/threats/sensitive-role-granted-by-ai-agent.md.txt)
- [`Persistence: Sensitive Role Granted to External AI Agent`](https://docs.cloud.google.com/security-command-center/docs/findings/threats/external-ai-agent-granted-sensitive-role.md.txt)
- [`Privilege Escalation: AI Agent Cross-Project Access Token Generation`](https://docs.cloud.google.com/security-command-center/docs/findings/threats/ai-agent-token-generation-cross-project-access-token.md.txt)
- [`Privilege Escalation: AI Agent Cross-Project OpenID Token Generation`](https://docs.cloud.google.com/security-command-center/docs/findings/threats/ai-agent-token-generation-cross-project-openid.md.txt)
- [`Privilege Escalation: AI Agent Token Generation Using Implicit Delegation`](https://docs.cloud.google.com/security-command-center/docs/findings/threats/ai-agent-token-generation-implicit-delegation.md.txt)
- [`Privilege Escalation: AI Agent Token Generation Using signJwt`](https://docs.cloud.google.com/security-command-center/docs/findings/threats/ai-agent-token-generation-sign-jwt.md.txt)

<br />

## What's next

- Learn about [Event Threat Detection](https://docs.cloud.google.com/security-command-center/docs/concepts-event-threat-detection-overview.md.txt).
- Learn about [Agent Platform Threat Detection](https://docs.cloud.google.com/security-command-center/docs/agent-engine-threat-detection-overview).
- Learn how to [respond to AI threat
  findings](https://docs.cloud.google.com/security-command-center/docs/respond-ai-threats).
- Refer to the [Threat findings index](https://docs.cloud.google.com/security-command-center/docs/threat-findings-index.md.txt).