How PPN 017 affects transport bids: AI transparency in central government procurement

PPN 017 does not ban AI in bids, and it does not bind local or combined transport authorities. It lets central government buyers ask where AI was used in a bid and check that what you submitted is accurate. The disclosure answer is not scored. That is the whole policy. What follows is how it lands on a transport bid team.

The people I speak to daily and the bids I engage with daily all have elements of machine assistance somewhere in the creation process. Government policy has caught on in a way to at least ask for that to be declared. However, most transport bid libraries have not caught up with the policy. They still cite the note that was replaced in February 2025. This page is the practical map. Which note applies to which tender, what a buyer can actually ask, and the one thing that I believe the policy is silent on.

1. What PPN 017 is

Procurement Policy Note 017, “Improving transparency of AI use in procurement”, was published by the Cabinet Office on 17 February 2025. It replaces PPN 02/24 for procurements commenced on or after 24 February 2025, the day the Procurement Act 2023 went live. It applies to all central government departments, their executive agencies and non-departmental public bodies. Other public sector contracting authorities “may wish to apply the approach”, which is the policy’s own wording for voluntary.

The note itself says the update “does not constitute a change in policy or a new call for action”. The substance is the same as PPN 02/24; the terminology has been updated to match the Procurement Act. So the question for a bid team is not “what changed” but “which note governs the tender in front of me”.

2. Which PPN applies to your tender, 02/24 or 017?

Whilst basic, two points make the table usable. First, a procurement is treated as commenced under the old rules where the contract notice was submitted for publication before the go-live date, so the tender notice tells you which regime you are in. Second, call-offs under a framework agreement or dynamic purchasing system that was concluded or established before 24 February 2025 continue under the Public Contracts Regulations 2015 for the life of that framework. For those tenders, PPN 02/24 remains the correct reference, and a large share of local authority and central purchasing body work will run on pre-Act frameworks for years yet.

The staleness test is therefore narrow and sharp. If your standard method statements, AI policy or win-theme boilerplate cite “PPN 02/24” on a tender that was commenced under the Procurement Act 2023, that is a visible signal to any procurement-literate evaluator that the library has not been read since early 2025.

Compliance is key. We have picked up these inconsistencies in multiple bid libraries, where the regulatory references were three regimes old and nobody had noticed, because the paragraph had scored well once and was never read again.

3. What PPN 017 asks authorities to do

Six things, taken from the note itself. Each one tells a bidder what to expect.

1. Treat supplier use of AI as permitted. The note states that “suppliers’ use of AI is not prohibited during the commercial process”.
2. Ask suppliers to disclose AI use in creating the tender response, using the example questions in Annex B. Those questions “should not be scored or taken into account when assessing a tender”.
3. Run proportionate due diligence where AI was used, which the note says “could include site visits, clarification questions or supplier presentations”, to check accuracy, robustness and credibility.
4. Put in place “proportionate controls to ensure suppliers do not use confidential contracting authority information” as training data for AI systems.
5. Consider “allowing more time in the procurement to allow for due diligence”, given higher response volumes.
6. Ask separately whether AI is part of the delivered service, not just the bid, where the service is not an AI service on its face.

4. Does PPN 017 apply to local transport authorities?

No. PPN 017 binds central government departments, their executive agencies and non-departmental public bodies. In transport that means the Department for Transport and its executive agencies directly. In practice DfT’s arm’s length bodies, including Network Rail and the DfT Operator train companies, apply Cabinet Office procurement policy too, so check the ITT. Transport for London is a functional body of the Greater London Authority and is outside the note’s scope. Local transport authorities, combined authorities and their bus franchising programmes are outside it too.

So in the part of the sector letting the largest bus contracts of the decade, AI transparency in bidding is currently voluntary. An authority that wants to can adopt the disclosure question, and publish a one-page statement of how it will and will not use AI in the process, at the cost of one page in the ITT.

With the enormous amount of focus on AI transparency, disclosure and general governance, my expectation is that the elements of PPN 017 will reach local and combined authorities quickly, whether through franchising guidance from the Department, through authorities’ own contract procedure rules, or simply because bidders start asking. Whoever takes the lead now sets the standard. AI is here whether we like it or not. It is almost certainly already used in every bid you receive, and we would be naive to think it isn’t, so think about it, have a policy on it, and have a process for dealing with it.

5. What does PPN 017 not say?

It does not address contracting authorities using AI to evaluate bids. The note points authorities to the government’s general AI guidance for their own use, and that guidance does not deal with tender evaluation. Nothing in PPN 017 tells you whether the person reading your method statement is a person.

It also carries no numbers. There is no count of how many bids are now AI-written and no measure of how often the disclosure question is asked. The one industry data point is Tussell’s reading, from January 2026, that “AI-generated content does not affect scoring, and fluent writing alone does not improve outcomes”, which is an assertion about scoring, not a statistic about bids. Nobody is counting, but I would place a bet that nearly every bid now uses AI somewhere, and that almost no ITT asks about it.

6. What should a transport bid team do now?

Bid teams:

1. Answer the disclosure question honestly, and verify every fact, case study and reference a model helped draft before submission. Under the Procurement Act 2023, “providing information that is incomplete, inaccurate or misleading” can count as acting improperly in a procurement, a discretionary exclusion ground where it gave you an unfair advantage (Schedule 7, paragraph 13). The due diligence in PPN 017 is aimed at exactly this.
2. Update the bid library. Check every PPN 02/24 reference against the rule in section 2: the note is right on pre-Act tenders and pre-Act frameworks, and wrong on everything commenced under the Act.
3. Write the one-page AI-use statement now, before anyone asks: what you use AI for in bid production, what stays human, and how confidential authority material is handled. Circulate it to everyone who takes part in writing a bid, from commercial to engineering to consultants; they all form part of this.
4. Keep a per-bid AI-use register, so a disclosure question never triggers an archaeology dig. You can even use AI to track this and keep the log.
5. Put authority documents behind your confidentiality controls. Do not upload ITT packs to tools that train on inputs, and read the ITT’s own AI and confidentiality clauses, which increasingly say so regardless of the PPN.
6. Expect the clarification stage to probe AI-assisted answers. Keep the evidence trail for every claim a model helped write.

Authorities:

7. Adopt the disclosure question voluntarily and say how you will use AI yourselves. The fairest competitions will be the ones that say so out loud.

For the bidder’s side of the machine first pass, see can AI increase your chance of winning a bid.

Related: AI use cases for UK public transport, six moves for authorities and operators in bus franchising, and the independent advisor agent.

This page is a policy summary written for bid teams, not legal advice.

Benjamin Ross. Practical application of AI in public transport. Founder, economAIcs. Benjamin Ross is the founder of economAIcs, the company behind YourTransportBrain.