The DICT-CSC Circular as a Procurement Condition

Who this is for: Owners, country managers, and sales leads at Japanese companies that already work with Philippine government agencies, or are considering it

Time required: 15 minutes

What this case study covers: How an AI framework issued for the Philippine public sector reaches the private companies that supply it. This is not a story about internal controls — it is about conditions for winning work.

Part 1: Read → Consider What It Means for Your Company

Here is what was issued.

Item	Detail
Issued	9 June 2026, Joint Memorandum Circular No. 003 by DICT and the Civil Service Commission
Content	A principles-based framework for public sector AI, built on governance, accountability and data protection
Application	Addressed to government agencies, but agencies are expected to secure equivalent compliance from third-party providers
Deadline	No effectivity date or compliance deadline is stated

What Happened

On 9 June 2026, the Philippine Department of Information and Communications Technology (DICT) and the Civil Service Commission (CSC) issued Joint Memorandum Circular No. 003, Series of 2026.

The document sets out a principles-based framework for the ethical and responsible development, deployment, and use of AI in the public sector. Governance, accountability, and data protection are its pillars.

It applies to government agencies. However, according to law-firm commentary, agencies are expected to ensure that third-party providers meet the same standards. That is where private companies come in.

Note that the circular states no effective date or implementation deadline. It is the kind of document that takes concrete shape through practice after issuance.

What Is Actually at Stake

It is tempting to skip a document addressed to government agencies. In practice, it means three different things depending on where you sit.

Companies that already hold government work — The standard may be applied at contract renewal or in the next procurement round.

Companies planning to bid for public work — More items will be asked at the proposal stage. Firms that have not prepared will be filtered out there.

Companies with no government business — No direct effect. But requirements that become standard in the public sector typically filter down into large-enterprise procurement over a few years.

There is one more easily missed point. Even if you are not the direct supplier, the same questions reach you as a subcontractor. Expect a Japanese prime contractor to ask how your company handles AI.

What It Means for You

This document is more useful read as a preview of procurement requirements than as a legal obligation.

Being principles-based, it contains no detailed technical requirements. That also means each agency will interpret it into concrete items of its own. Companies that prepare their explanatory materials in advance will find those interpretations far easier to work with.

Part 2: Key Terms Explained (for Executives)

Principles-based framework — A document that states the thinking to be followed rather than specific figures or procedures. The absence of detailed rules means requirements vary by agency. Flexible, but hard to predict in advance.

Governance — The structure defining who decides, who approves, and who bears responsibility. What gets tested is whether you can name the individuals.

Accountability — Being able to explain what happened when something goes wrong. Without records, you cannot explain.

Third-party provider — A private business that supplies, develops, or operates AI systems for government agencies. Under this framework, agencies are expected to ensure such providers comply.

Data protection — In the Philippines, the Data Privacy Act (Republic Act No. 10173) is the baseline. For AI that handles personal information, that law and the National Privacy Commission's guidance take effect first.

Part 3: Applying It to Your Company

For a site handling government work, proceed in this order.

Step	What to do	The point
1	Inventory whether AI is in your deliverables	Use during development (code, documents, test data) is the most overlooked
2	Get to a state where you can explain it	Models and uses / who checks and approves / where logs live and for how long. One or two pages is enough
3	Write down where the data goes	Whether it leaves for servers outside the Philippines will always be asked
4	Decide where a human checks	Anything that leaves the company, cannot be undone, or ends up on the record goes through a person
5	Review the contract clauses	Whether prior notice or approval is required for AI use

At any office handling — or planning to handle — government work, proceed in this order.

Step 1: Inventory Whether AI Is Part of What You Deliver

Companies that assume "we don't supply AI" often find it is there after all.

Does the system you deliver include summarisation, classification, translation, or chat response?
Do you use generative AI during development (code, documentation, test data)?
Do you use AI in operations or maintenance?

Use during development is the most overlooked. Even when the deliverable itself contains no AI, you may be asked how it was built.

Step 2: Get to a State Where You Can Explain

Prepare the material you would hand over if asked. At minimum, three points:

Which model, for which purpose
Who reviews the output, and who approves it
Where records are kept, and for how long

One or two pages is enough. Having it at all is the dividing line; precision can be added later.

Step 3: Write Out Where the Data Goes

Diagram where government data travels. If you use an external AI service, that data leaves the Philippines for servers abroad.

You will be asked about this. If you cannot answer, the decision stalls on the spot.

Step 4: Decide Where a Human Checks

Identify any point where AI output becomes part of the deliverable directly, and insert human review there.

The test is simple: anything that goes outside, cannot be undone, or is recorded goes through a person. Internal drafts and in-house summaries can move freely.

Step 5: Review Your Contract Clauses

For existing government work, check whether these three points appear in the contract:

Whether prior notice or approval is required for AI use
Record retention periods, and handling when disclosure is requested
What happens if the provider's service changes or is discontinued

Where they are absent, the other side will add them at the next renewal. Proposing them yourself lets you shape the terms.

Across my development career I have insisted on detailed written specifications and mandatory multi-person review. However strong the technology, a system only one person understands weakens the whole business. In public-sector work, that documentation becomes proposal material directly.

Part 4: Common Failure Patterns (What Not to Do)

NG1: Skipping it because "it's for government"

The circular applies to agencies, but agencies are expected to ensure third-party providers meet the same standards. The structure passes through to suppliers, so ignoring it puts you at a disadvantage in the next procurement.

NG2: Reading "no effective date" as "not urgent"

The circular sets no deadline. That does not mean you have time. It can be raised at any agency's procurement stage, and requirements without deadlines tend to arrive without warning.

NG3: Leaving it to the technical team

What gets asked about is governance and accountability, not model performance. Since the questions concern who approves and who bears responsibility, they cannot be answered without management and legal involvement. Hand it to engineering and the answers come back in the wrong shape.

NG4: Submitting the Japanese head office policy unchanged

Even if your parent company has an AI usage policy, do not submit it as is. It was not written against the Philippine Data Privacy Act or this circular. It needs translating into the local context, not just into English.

NG5: Not telling local staff

When the practice described in your proposal differs from how the team actually works, the gap surfaces during review. In the Philippines a casual conversation is often treated as an effective agreement, so a change in the scope of AI use needs to reach the agenda explicitly, not just the corridor.

Practical Tips (3 Tips)

Tip 1: Try it on one engagement — Drafting a company-wide policy never finishes. Pick one live government project and build the explanatory material for that one. Once you have one, the rest is duplication.

Tip 2: Write down what you are not doing — For items you will not address, record the reason and when you will revisit them. A blank leaves the reviewer unable to judge.

Tip 3: Cite the source when persuading internally — Grounding the case in a document from a public agency makes it easier to explain to the Japanese head office, and easier to get budget approved.

Bonus: How to Make Use of PH AI Works

We can review how to build the AI explanatory materials that Philippine public-sector work now asks for, and which clauses in your existing contracts need revisiting, against the specifics of your engagements. We can also start from an inventory of systems you have already delivered.