EU AI Act data governance framework for enterprise LLM deployment — Reference Architecture

Reference architecture · published 2026-10-16 · regulated enterprise · EU AI Act · layout diagram_first · implementation executed in CI before publish

The control that resolves this is a Classification-Routed Multi-Cloud Estate, evaluated before the request reaches a model provider rather than audited afterwards. In regulated enterprise the binding constraint is EU AI Act, and the question it actually asks is what risk class is this system, and where is the technical documentation. The shape below holds p95 ≤ 600ms over REST over HTTPS, and ships with a standard-library implementation that runs as written.

Reference architecture

Classification-Routed Multi-Cloud Estate — the request path, with the decision points that must exist before the first model call:

flowchart TB
    IN[Request payload] --> CLS[Deterministic classifier<br/>regex + Luhn/checksum]
    CLS --> LV{Classification}
    LV -->|PUBLIC / INTERNAL| D1[shared-us]
    LV -->|CONFIDENTIAL| D2[eu-pooled<br/>DPA required]
    LV -->|RESTRICTED| D3[eu-dedicated<br/>zero-retention + subprocessors]
    LV -->|no lawful destination| DN[Deny + reason]
    D1 --> LOG[(Decision log)]
    D2 --> LOG
    D3 --> LOG
    DN --> LOG
    D3 --> LIN[Lineage tag on<br/>embeddings / cache / evals]
    ATT[Destination attestations] -.-> LV
    ATT --> DR[Drift detector]

The parts you have to build

1. Deterministic classifier — Pattern plus validator (Luhn, checksum) so the gate does not cry wolf.
2. Destination attestations — Region, max class, retention, DPA, subprocessor disclosure — as data.
3. Routing gate — First lawful destination in preference order; cost can never override a control.
4. Decision log — A recorded reason for every allow AND every deny.
5. Lineage tags — Classification travels with derived artifacts: embeddings, caches, evals.
6. Drift detector — Alerts when a destination's attested properties change.

Deployment parameters

Every value below is a parameter you can hold a vendor or an internal team to. Adjectives are not acceptance criteria.

Where the rollout actually stalls

Workloads span several providers and regions, and the decision of which model may see which data lives in prompt text, a wiki page, or a developer's judgement. That is not a control, so the governance board cannot approve the estate even though every individual component is defensible.

What EU AI Act specifically requires

The EU AI Act regulates the SYSTEM, not the data. The first question is its risk classification, and a high-risk classification pulls in Article 11 technical documentation, Article 12 record-keeping and Article 14 human oversight — obligations that have to be designed in, not attached later.

The artifact that satisfies it is a model card plus Article 12 automatic logs. The way teams most often fail it: a system classified as limited-risk that is actually high-risk because it informs a decision about a person.

Reference implementation — Classification-driven routing gate with recorded reasons

Standard-library Python, no external dependencies, runs as written. Executed by the publish gate: if it raises, this page does not ship. Wire it behind REST over HTTPS; the failure mode it guards against is a retried POST creating a duplicate record. The output you hand the assessor is the Annex III classification record and model card.

"""Classification-driven routing gate for a multi-cloud LLM estate.

Why this shape: "which model may see this data" is a governance decision, and
if it lives in prompt text or a developer's judgement it is not a control. It
has to be a deterministic function of the data's classification and the
destination's attested properties, evaluated before egress.

The gate below is the smallest thing that an actual data-governance board will
accept: classification in, allowed destinations out, with a recorded reason for
every allow and every deny.
"""
import re
from dataclasses import dataclass
from enum import IntEnum


class Class(IntEnum):
    PUBLIC = 0
    INTERNAL = 1
    CONFIDENTIAL = 2
    RESTRICTED = 3      # PII / PHI / material non-public


DETECTORS = {
    "email": re.compile(r"[\w.+-]+@[\w-]+\.[\w.]{2,}"),
    "iban": re.compile(r"\b[A-Z]{2}\d{2}[A-Z0-9]{10,30}\b"),
    "card": re.compile(r"\b(?:\d[ -]?){13,19}\b"),
    "ssn": re.compile(r"\b\d{3}-\d{2}-\d{4}\b"),
    "mrn": re.compile(r"\bMRN[:\s-]*\d{6,10}\b", re.I),
}


def luhn(number: str) -> bool:
    """Card detectors without Luhn produce false positives on order numbers,
    and a gate that cries wolf gets switched off."""
    digits = [int(c) for c in re.sub(r"\D", "", number)]
    if len(digits) < 13:
        return False
    checksum, parity = 0, len(digits) % 2
    for i, d in enumerate(digits):
        if i % 2 == parity:
            d *= 2
            if d > 9:
                d -= 9
        checksum += d
    return checksum % 10 == 0


def classify(text: str, floor: Class = Class.INTERNAL) -> tuple:
    """-> (classification, evidence). Deterministic and explainable."""
    found = []
    for name, rx in DETECTORS.items():
        for m in rx.finditer(text or ""):
            if name == "card" and not luhn(m.group(0)):
                continue
            found.append(name)
            break
    return (Class.RESTRICTED if found else floor), sorted(set(found))


@dataclass(frozen=True)
class Destination:
    name: str
    region: str
    max_class: Class
    zero_retention: bool
    dpa_signed: bool
    subprocessors_disclosed: bool


class Denied(Exception):
    pass


def route(text: str, destinations: list, required_region: str = None) -> dict:
    """Pick the first destination whose ATTESTED properties cover the data.

    Order matters: destinations are supplied in preference order (cheapest or
    fastest first) and the gate takes the first one that is lawful, so cost
    optimisation can never override a control.
    """
    level, evidence = classify(text)
    reasons = []
    for d in destinations:
        if d.max_class < level:
            reasons.append(f"{d.name}: max_class {d.max_class.name} < {level.name}")
            continue
        if required_region and d.region != required_region:
            reasons.append(f"{d.name}: region {d.region} != {required_region}")
            continue
        if level >= Class.RESTRICTED and not d.zero_retention:
            reasons.append(f"{d.name}: retains prompts, unusable for {level.name}")
            continue
        if level >= Class.CONFIDENTIAL and not d.dpa_signed:
            reasons.append(f"{d.name}: no DPA on file")
            continue
        if level >= Class.RESTRICTED and not d.subprocessors_disclosed:
            reasons.append(f"{d.name}: subprocessors undisclosed")
            continue
        return {"allowed": True, "destination": d.name,
                "classification": level.name, "evidence": evidence,
                "rejected": reasons}
    raise Denied(f"no lawful destination for {level.name} data: {reasons}")


if __name__ == "__main__":
    dests = [
        Destination("shared-us", "us-east-1", Class.INTERNAL, False, True, False),
        Destination("eu-pooled", "eu-west-1", Class.CONFIDENTIAL, False, True, True),
        Destination("eu-dedicated", "eu-west-1", Class.RESTRICTED, True, True, True),
    ]
    print(route("Summarise this quarter's churn drivers.", dests))
    print(route("Patient MRN 4471902 reports side effects.", dests,
                required_region="eu-west-1"))
    try:
        route("Card 4111111111111111 was declined.",
              [d for d in dests if not d.zero_retention])
    except Denied as e:
        print("correctly denied:", str(e)[:96])

The EU AI Act control, as code

Annex IV documentation completeness and Article 12 logging. This is the artifact an assessor asks for under EU AI Act specifically — it is not the same control as the one above, and a system that implements one does not thereby implement the other. It executes under the same publish gate.

"""Annex IV technical documentation completeness + Article 12 logging.

A high-risk system under the EU AI Act needs technical documentation covering
specified elements, and automatic recording of events over the system lifetime.
The failure is rarely refusal to document — it is documentation that is stale,
partial, or cannot be tied to the exact model version that produced an output.

This checks the documentation set against the Annex IV elements and refuses to
mark a release conformant while any element is missing or stale.
"""
from datetime import date, timedelta

ANNEX_IV_ELEMENTS = {
    "general_description": "intended purpose, provider, version",
    "design_specification": "architecture, algorithmic logic, design choices",
    "training_data": "provenance, scope, characteristics, labelling",
    "human_oversight": "measures enabling natural-person oversight",
    "accuracy_metrics": "accuracy, robustness and cybersecurity metrics",
    "risk_management": "risk management system per Article 9",
    "change_log": "changes made through the lifecycle",
    "post_market_plan": "post-market monitoring plan",
}

STALENESS_DAYS = 180


class TechnicalFile:
    def __init__(self, system: str, model_version: str, today: date):
        self.system, self.model_version, self.today = system, model_version, today
        self.elements: dict[str, date] = {}

    def attach(self, element: str, updated: date) -> None:
        if element not in ANNEX_IV_ELEMENTS:
            raise KeyError(f"{element} is not an Annex IV element")
        self.elements[element] = updated

    def missing(self) -> list[str]:
        return sorted(set(ANNEX_IV_ELEMENTS) - set(self.elements))

    def stale(self) -> list[str]:
        cutoff = self.today - timedelta(days=STALENESS_DAYS)
        return sorted(k for k, v in self.elements.items() if v < cutoff)

    def conformant(self) -> dict:
        missing, stale = self.missing(), self.stale()
        return {"system": self.system, "model_version": self.model_version,
                "missing": missing, "stale": stale,
                "conformant": not missing and not stale}


class EventLog:
    """Article 12: records must tie an output to the version that produced it."""

    def __init__(self):
        self.events: list[dict] = []

    def record(self, model_version: str, event: str, ref: str) -> None:
        self.events.append({"model_version": model_version,
                            "event": event, "ref": ref})

    def trace(self, ref: str) -> dict | None:
        for e in self.events:
            if e["ref"] == ref:
                return e
        return None

    def untraceable(self, known_versions: set[str]) -> list[str]:
        return [e["ref"] for e in self.events
                if e["model_version"] not in known_versions]


if __name__ == "__main__":
    today = date(2026, 10, 16)
    tf = TechnicalFile("claims-triage", "v4.2.0", today)
    for element in ANNEX_IV_ELEMENTS:
        tf.attach(element, today - timedelta(days=30))
    assert tf.conformant()["conformant"], tf.conformant()

    partial = TechnicalFile("claims-triage", "v4.3.0", today)
    for element in list(ANNEX_IV_ELEMENTS)[:5]:
        partial.attach(element, today - timedelta(days=10))
    partial.attach("change_log", today - timedelta(days=400))
    report = partial.conformant()
    assert not report["conformant"], report
    assert "post_market_plan" in report["missing"], report
    assert "change_log" in report["stale"], report

    log = EventLog()
    log.record("v4.2.0", "inference", "out-9001")
    log.record("v4.1.7", "inference", "out-8800")
    assert log.trace("out-9001")["model_version"] == "v4.2.0"
    assert log.untraceable({"v4.2.0"}) == ["out-8800"]

    print("EU AI Act Annex IV + Article 12: OK")
    print(f"  complete file conformant={tf.conformant()['conformant']}, "
          f"partial missing {len(report['missing'])}, stale {len(report['stale'])}")

Risk register with acceptance criteria

Each row carries an acceptance parameter rather than an adjective, because "low latency" cannot be signed off. Retention here is bound by per the data class retention schedule, enforced automatically.

Review questions and answers

• What is this system's risk classification under Annex III, who decided, and what happens to that classification when the use case changes?
  The classification is recorded against its Annex III category with a named owner, and a material change of use triggers re-classification before the change ships.
• What function decides which destination may see RESTRICTED data, and where does it run?
  The routing gate, evaluated before egress. It is a deterministic function of the data's classification and the destination's attested properties, and it fails closed when no destination is lawful.
• Show me the recorded reason for a denial last week.
  Every allow and every deny is written to the decision log with the reason — which destinations were rejected and on which property — so a denial is explainable months later.
• When a document is embedded, does the embedding carry its classification — and is that enforced on read?
  Yes. The classification tag propagates into every derived store, including embeddings, caches and eval sets, and is enforced on read as well as write.
• What happens when a provider adds a subprocessor you have not disclosed?
  The drift detector alerts when a destination's attested properties change, and routing is re-evaluated against the new attestation. Attestations are versioned data, not a one-off questionnaire.
• How do you prove processing stayed in the contracted region?
  The provider endpoint is pinned per region and the served region is asserted on the response; a mismatch fails the request rather than completing it quietly.

See also

• Part of enterprise ai data governance.
• For the related constraint, see SOX data governance framework for enterprise LLM deployment.
• fedRAMP Moderate data governance framework for enterprise LLM deployment applies the same mechanism under a different obligation.
• This pattern also governs PCI DSS 4.0 data governance framework for enterprise LLM deployment.
• Where this interacts with another regime, see ISO/IEC 27001:2022 data governance framework for enterprise LLM deployment.
• If this is the blocker in front of you: enterprise AI infrastructure review.