Query Public AI Services

Executive Summary

Query Public AI Services (T1682) is a MITRE ATT&CK technique associated with Reconnaissance. Adversaries may query publicly accessible artificial intelligence (AI) services, such as large language models (LLMs), to support targeting and operations.

Why Attackers Use It

Attackers use Query Public AI Services because it provides a reliable way to advance their objective within the Reconnaissance tactic, often with a favorable balance of impact versus detectability on PRE environments. Defenders should assess this behavior in the context of the affected platform and adjacent activity rather than treating it as a standalone indicator.

MITRE Description

Adversaries may query publicly accessible artificial intelligence (AI) services, such as large language models (LLMs), to support targeting and operations. In addition to searching websites or databases directly (i.e., Search Open Websites/Domains), adversaries may use AI services to synthesize, aggregate, and analyze publicly available information at scale. This may include identifying individuals or organizations to target, researching organizational structures and personnel, identifying technologies used by target organizations, researching business relationships to develop plausible pretexts for Social Engineering approaches, identifying contact information for use in Phishing or Phishing for Information, or gathering derogatory or sensitive information about individuals that may be used for extortion or coercion.(Citation: MSFT-AI)(Citation: GTIG AI Threat Tracker)

Information gathered through AI services may be leveraged for other behaviors, such as establishing operational resources (i.e., Generate Content or Establish Accounts. For obtaining access to AI tools and services, see Artificial Intelligence.

Attack Flow

1. Attacker gains the prerequisite access or context described below.
2. Attacker executes Query Public AI Services to achieve its tactical objective (Reconnaissance).
3. Resulting access/data/effect is leveraged to advance the broader attack chain (see Related Techniques).

Prerequisites

• Platform(s): PRE
• ATT&CK does not define one universal permission requirement for this technique. Establish the required access from the observed implementation and affected platform.

Common Tools

• Tool attribution is implementation-specific. Use ATT&CK procedure examples and local telemetry to identify the binaries, services, scripts, accounts, or cloud resources involved.

Commands

No universal command represents Query Public AI Services. Capture the exact command line, arguments, parent process, account, host, and execution time from the investigated environment; do not operationalize unverified examples.

Network Traffic

• Network observability is implementation-dependent. Review DNS, proxy, firewall, flow, authentication, and packet telemetry around the activity window, then correlate remote endpoints and protocol behavior with host evidence.

Windows Events

Sysmon Events

Detection Opportunities

No MITRE detection guidance published for this technique.

Relevant ATT&CK Data Sources: N/A

Sigma Rules

A universal Sigma rule would create unreliable results because this technique has no single guaranteed observable. Build detection logic from a documented behavior and supported data source, scope it to the affected platform, and validate it against benign administrative activity before deployment.

Splunk Queries

Start with the data sources named in the detection section. Scope searches by asset, identity, and time window; correlate the primary behavior with preceding access and subsequent actions. A portable query is intentionally not provided where the technique lacks a universal schema or observable.

Investigation Workflow

1. Confirm that the observed behavior is consistent with Query Public AI Services and rule out expected administrative or application activity.
2. Establish the first-seen time, initiating identity, source system, target system, and affected resources.
3. Collect relevant host, identity, network, cloud, and application telemetry for the surrounding time window.
4. Correlate parent and child activity, remote connections, file or configuration changes, and related ATT&CK techniques.
5. Determine scope by searching for the same observable across peer assets and identities.
6. Preserve volatile evidence and record confidence, assumptions, and telemetry gaps before containment.

Containment

1. Isolate affected host(s)/account(s) identified during investigation.
2. Revoke or rotate any credentials/tokens potentially exposed.
3. Apply the mitigations listed below where not already enforced.
4. Validate no related techniques (see Related Techniques) were chained against the same asset.

Mitigation

• M1056 -- Pre-compromise: Pre-compromise mitigations involve proactive measures and defenses implemented to prevent adversaries from successfully identifying and exploiting weaknesses during the Reconnaissance and Resource Development phases of an attack.

Related Techniques

No related techniques mapped.