Governance / Comparisons

FERZ is not SIEM

FERZ is a runtime AI-action authorization architecture with built-in action canonicalization, governed state, ABSTAIN-with-override semantics, and a proof-carrying authorization artifact. Security Information and Event Management (SIEM) is a class of platforms that aggregate security events from enterprise systems, apply correlation rules, and generate alerts for human investigators. SIEM is necessary security infrastructure; it operates downstream of action execution and does not constitute AI-action authorization.

This comparison addresses a common procurement pattern: an enterprise extends its existing SIEM to ingest AI-system events and treats the resulting dashboards as AI governance. The two architectures operate at fundamentally different layers. SIEM is post-hoc and observability-class: it sees actions after they have executed and surfaces patterns for human investigators. FERZ is ex-ante and authorization-class: it evaluates proposed actions against policy before execution and emits authorization verdicts. The architectural-class test is established in The Authorization Artifact Test (Meyman, 2026; DOI 10.5281/zenodo.20013582); the observability-vs-authorization distinction is developed in Observability Is Not Enforcement (Meyman, 2026; DOI 10.5281/zenodo.18663864) and the impossibility of deriving authorization from observation in On the Impossibility of Observability-Based Authorization (Meyman, 2026; DOI 10.5281/zenodo.19647542).

Where the categories diverge

The two architectures diverge along four architectural axes.

Execution locus
FERZ

At the AI action proposal boundary. The authorization verdict precedes execution of the proposed action, with built-in action canonicalization and AI-action semantics.

SIEM

At the event ingestion boundary, after execution. The SIEM receives events emitted by source systems; by the time SIEM sees an action, the action has already happened.

Authority source
FERZ

Encoded policy evaluated against a canonicalized action specification, governed state, and policy version. The question asked is whether the action is authorized.

SIEM

Correlation rules applied to event streams across sources. The question asked is whether this pattern of events matches an alerting condition.

Evidence artifact
FERZ

A verdict in the canonical verdict space {ALLOW, DENY, ABSTAIN} with an authorization artifact reconstructable from policy, action specification, and governed state.

SIEM

Alerts, incident records, correlation outputs. The artifact describes what was observed across event streams and what should be investigated; it is not a verdict on a proposed action.

Default execution behavior
FERZ

Fail-closed by architectural design. Execution is blocked when governance conditions are not met.

SIEM

Alerts-only by design. SIEM does not prevent actions; it surfaces them for human investigation. Response is mediated by SOAR playbooks, runbooks, or human operators.

Why the distinction is structural

SIEM is post-hoc; FERZ is ex-ante. A SIEM platform ingests events emitted by source systems after those systems have acted: log lines, API call records, authentication events, system telemetry. The SIEM aggregates these events across the enterprise, applies correlation rules, and surfaces patterns for human investigators. By the time the SIEM sees an action, the action has already executed and produced whatever consequence it produced. FERZ evaluates proposed actions before they execute. The architectural layer is fundamentally different: observability after the fact versus authorization before the fact.

The authority source is correlation, not authorization. A SIEM rule asks whether this pattern of events across event streams matches an alerting condition. The output is a confidence-weighted alert: this pattern looks unusual, this pattern matches a known indicator of compromise, this pattern correlates across systems in a way that warrants human attention. FERZ asks whether a specific proposed action is permitted by policy. The output is a verdict reconstructable from policy, action specification, and governed state. Correlation outputs and authorization verdicts are not points on a spectrum; they answer different questions about different objects.

The evidence artifacts reflect this. A SIEM artifact describes what was observed and what should be investigated: an alert with severity, correlation source, time window, and recommended response. The reader is a human investigator deciding what to look into. A FERZ verdict describes what was permitted: policy version, canonicalized action specification, governed state, evaluator version, and the resulting decision. The reader is a third party reconstructing the decision against the same policy and state. Alerts and authorization verdicts serve different audiences and answer different questions; a more sophisticated correlation engine produces better alerts, not authorization properties.

The structural claim follows. SIEM and FERZ are layered, not substitutable. A mature AI deployment has both: FERZ at the action proposal boundary authorizing individual proposed actions; SIEM ingesting the resulting verdicts, overrides, and operational events alongside the rest of the enterprise security event stream for correlation and incident response. SIEM ingesting FERZ events does not produce AI-action authorization; producing authorization is FERZ's job, and the SIEM treats those authorization events as one signal source among many. The architectural class remains post-hoc event aggregation regardless of which events are being ingested. Calling a SIEM module "AI governance" in the authorization sense is a category error that the impossibility result establishes formally.

What buyers often confuse

The most common conflations involve enterprise security platforms extended to cover AI workloads and positioned as governance. The architectural argument applies regardless of presentation.

SIEM platforms extended with AI-related event sources.A SIEM such as Splunk, Microsoft Sentinel, or IBM QRadar configured to ingest events from AI systems alongside other enterprise sources. The platform correlates events post-hoc and surfaces alerts; it does not evaluate proposed actions against governing policy before execution.
AI-specific "security analytics" modules.Vendor modules that apply correlation rules to AI-related event streams and produce AI activity dashboards. The system detects patterns in observed activity; it does not authorize proposed actions, and the dashboard is descriptive, not decisional.
Incident response playbooks for AI workloads.SOAR-class workflows triggered by SIEM alerts on AI-related activity. The playbook responds to observed events with automated or human-driven remediation; it does not authorize what may occur. Response and authorization are different operations.
Compliance dashboards built on SIEM.Dashboards that aggregate AI-system events for compliance reporting, sometimes labeled as AI governance. The dashboard reports what happened across event streams; it does not authorize proposed actions or produce policy-versioned authorization artifacts.

Architectural diagnostic

Any vendor or product claiming AI governance can be evaluated against four architectural questions, anchored to The Authorization Artifact Test (Meyman, 2026; DOI 10.5281/zenodo.20013582):

1
Pre-execution artifact

Does the system produce an artifact for the proposed AI action that exists before the action executes?

2
Reconstructable decision

Is the authorization artifact reproducible from policy, canonicalized action specification, and governed state, sufficient for independent third-party verification?

3
AI-specific verdict semantics

Does the system include an ABSTAIN verdict with explicit fail-closed semantics pending authorized human override?

4
Default failure behavior

Is fail-closed the architectural default, or an opt-in configuration?

FERZ answers all four questions affirmatively by architectural commitment. Any architecture claiming AI governance that fails one or more of these is operating in an adjacent category, not in AI-action authorization.

References

Meyman, E. (2026). The Authorization Artifact Test: Applying the Impossibility Result to Ex-Ante Regulatory Regimes. Zenodo. https://doi.org/10.5281/zenodo.20013582
Meyman, E. (2026). A Taxonomy of AI Governance Approaches: Distinguishing Visibility, Alignment, and Authorization. Zenodo. https://doi.org/10.5281/zenodo.18275969
Meyman, E. (2026). On the Impossibility of Observability-Based Authorization: A Formal Impossibility Result for Ex-Ante AI Governance. Zenodo. https://doi.org/10.5281/zenodo.19647542
Meyman, E. (2026). Observability Is Not Enforcement: A Doctrinal Framework for Distinguishing Compliance Instrumentation from Runtime Authorization in AI Governance Architectures. Zenodo. https://doi.org/10.5281/zenodo.18663864
Meyman, E. (2026). From Monitoring to Authorization: The Structural Shift in Agentic AI Governance. Zenodo. https://doi.org/10.5281/zenodo.18743974

Related doctrine and concepts

Frequently asked questions

Our SIEM ingests events from our AI systems. Doesn't that handle AI governance?

SIEM ingests events after they occur. By the time SIEM sees an AI action, the action has already executed and produced whatever consequence it produced. SIEM can surface patterns of AI activity for human investigators and trigger response playbooks. It does not authorize proposed AI actions before they execute. The architectural layer is post-hoc observability, not ex-ante authorization.

Can SIEM detect AI-action policy violations?

SIEM can be configured to alert on patterns that may indicate policy violations: unusual action frequency, anomalous parameter ranges, correlations across event streams. These are post-hoc detections of patterns that look like violations. They are not policy-versioned authorization verdicts on individual proposed actions; they are correlation results that flag candidates for human investigation.

How does FERZ relate to our SIEM and SOAR investments?

FERZ and SIEM operate at different architectural layers and compose well. FERZ authorizes AI actions before they execute, producing verdicts and authorization artifacts. SIEM ingests events from across the enterprise, including FERZ verdicts and overrides, for correlation and incident response. The SIEM surfaces patterns; FERZ decides individual actions. Both have roles in a mature AI deployment.

Vendors describe SIEM modules as "AI governance compliance." Is that accurate?

Compliance reporting on observed AI activity is accurate. AI-action authorization is a category error when applied to SIEM, because SIEM does not evaluate proposed actions against policy before execution. A compliance dashboard built on SIEM tells you what happened; authorization tells you what may happen. The architectural layers are distinct.

Where does SIEM fit in a FERZ-governed AI stack?

SIEM ingests the FERZ event stream alongside other enterprise security events: verdicts emitted, overrides issued, policy versions in force, anomalies in override patterns. The SIEM treats FERZ as one signal source among many, applying correlation and alerting on patterns of governance activity. FERZ produces verdicts; SIEM produces situational awareness of how those verdicts are being issued and overridden.

Cite this page

FERZ, Inc. (2026). FERZ is not SIEM. https://ferz.ai/governance/comparisons/ferz-is-not-siem

BibTeX
@misc{ferz2026notsiem,
  author = {{FERZ, Inc.}},
  title  = {FERZ is not SIEM},
  year   = {2026},
  url    = {https://ferz.ai/governance/comparisons/ferz-is-not-siem},
  note   = {FERZ Architectural Comparison}
}