Is the ECC mandatory for us?

It is the national baseline for government entities and critical national infrastructure, and it is widely used as the reference by regulated private firms and by anyone bidding for government work.

Our security team owns this. Why involve the data team?

Because the evidence lives in the platform. Access reviews, encryption of sensitive columns and log enrichment cannot be produced from outside the warehouse.

How do LLM services fit the third party controls?

As external processors. What matters is the data class you send, where inference runs, whether prompts are retained, and whether the route is enforced in code rather than in a policy note.

Cybersecurity

Where the Essential Cybersecurity Controls meet your data and AI platform

Security teams own the ECC. Data teams keep discovering they own half of it too, because the controls land on identity, logging, encryption and the third parties a pipeline calls.

Book the Data Foundation Assessment AI governance practice

Most audit findings on a data platform are access findings. Shared accounts and forgotten service credentials, not exotic attacks.

NCA ECCSDAIAPDPL

The short answer

The National Cybersecurity Authority's Essential Cybersecurity Controls set a baseline for national organisations across governance, defence, resilience and third party security. On a data and AI platform the controls that bite are identity and access management, encryption in transit and at rest, event logging and monitoring, secure development, and cloud and third party assurance.

Key takeaways

What you need to know, in a minute

Identity is the perimeter

Named identities, least privilege and periodic review beat network controls once analysts and services both query the same tables.

Logs are only useful if they are queryable

Access events need to answer who read which classified table, when, and from where, without a week of log archaeology.

Third parties inherit your obligations

Every managed service, model API and offshore support path is in scope. The pipeline that calls it is the one that exposes you.

AI adds a new asset class

Prompts, embeddings, vector stores and model outputs carry the classification of the data behind them and need the same controls.

From the text to what gets built

Each line is a published expectation and what we implement against it. None of this is legal advice or a compliance verdict.

Area	What it asks for	What we build
Identity and access	Control and review who can reach what.	Role and attribute based access on the warehouse, service identities without shared secrets, and access reviews driven from the catalogue.
Cryptography	Protect data in transit and at rest.	Managed keys with rotation, column level protection on sensitive attributes, and no plaintext copies in staging areas.
Event logs and monitoring	Collect, protect and review security events.	Access and query logs shipped to the security platform with the data classification attached, so alerts carry meaning.
Secure development	Build and change systems securely.	Pipelines as reviewed code, secrets out of repositories, environments separated, and releases traceable to a change record.
Third party and cloud	Assure the security of external providers.	An inventory of every external service a pipeline or agent calls, what data class reaches it, and where it runs.
Resilience	Keep critical services recoverable.	Tested restore of the platform and its metadata, not only of the raw storage bucket.

Access model rebuild

Shared logins and inherited grants replaced with roles tied to the data classification and reviewed on a schedule.

Audit ready logging

Query and access events enriched with classification and owner, retained where the auditor can read them.

AI asset controls

Vector stores, prompts and model logs brought under the same classification, retention and access rules as the source data.

Provider inventory

One list of external dependencies with data class, hosting location and the contract position for each.

Use cases

Where this applies in practice

Questions we get on NCA Essential Cybersecurity Controls

Related guides

PDPL

Most gaps we find are not legal gaps. They are missing lineage, missing classification and copies nobody owns.

Open the guide

NDMO controls

An assessment score moves when the catalogue is populated by the pipeline, not by a spreadsheet campaign.

Open the guide

SDAIA alignment

The ethics principles are not decoration. They decide which decisions a model is allowed to make alone.

Open the guide

SAMA expectations

The question is rarely whether the model works. It is whether you can reconstruct one decision from eighteen months ago.

Open the guide

In-country hosting

The failure is almost never the main database. It is a notebook, a support tunnel or a model API nobody classified.

Open the guide

Etimad

A bid rarely loses on the model. It loses on the appendix nobody prepared.

Open the guide

Let's put your AI ambition on the table

Sixty minutes with a senior Bilytica partner. We stress-test your plans against what actually works in your sector, agree which moves are worth funding first, and you leave with a written 90-day plan and honest numbers, whether you work with us or not.

Start the conversation