The EU's new political ad repository goes live after regulation takes effect

A register anyone can search

The European Commission's transparency repository for political advertising went live following the entry into force of Regulation (EU) 2024/900 on 10 October 2025 ^[1]. The repository collects transparency notices that sponsors and platforms are required to publish, making it possible to search who paid for an ad, which election or referendum it relates to, how much was spent and whether targeting was used ^[2].

Think of it as a public library card catalogue, only for political ads instead of books. You can look up an ad and see who sponsored it, when it ran and what methods were used to decide who saw it. The repository itself doesn't decide whether an ad broke any rules; it gives people the information to ask the question.

The regulation defines political advertising broadly: ads placed by or on behalf of a political actor, or about an election, a referendum, a political party or an issue likely to influence the outcome of a vote ^[2]. An agency running a campaign for a party, a platform showing ads about a referendum, or an influencer posting sponsored content in the run-up to an election can all fall within the scope.

What sponsors and platforms must keep

Sponsors and providers of political advertising services must keep records of each campaign for seven years ^[2]. That includes the sponsor's identity, the amounts paid, which election or referendum the ad relates to and whether targeting methods were used. If the ad was targeted using personal data, the records must show the categories of data used and the criteria applied.

Platforms that carry political ads must ask the sponsor whether the service concerns political advertising and, if it does, retain the prescribed information ^[2]. The transparency notice must be accessible and machine-readable, which in practice means structured data that supervisors and researchers can query.

Fines for failing to comply can reach 6 per cent of the sponsor's or provider's annual income, budget or global turnover, whichever applies ^[2].

Online targeting: consent and no sensitive profiling

When political advertising online uses personal data for targeting, the data must have been collected from the person concerned, the person must have given explicit consent to receive political ads, and targeting may not rely on profiling using sensitive personal data ^[2]. The categories listed as sensitive include racial or ethnic origin, political opinions, religious beliefs, trade union membership, genetic data, biometric data used for identification, health data and data about sex life or sexual orientation.

That last point means a campaign can't build a profile from someone's health records, their union membership or inferred political leanings and then target ads based on that profile. It can use data the person has knowingly shared and consented to, such as signing up to a party mailing list, but not data pieced together from other sources without explicit consent.

A seven-year record you can prove

The seven-year retention period means a campaign archive that survives multiple elections and long after the people who ran the campaign have moved on. When a question comes up about who approved a version of an ad, whether targeting settings matched what the transparency notice said, or which influencer deals were agreed, the answer sits in records that may be scattered across agencies, platforms and internal systems.

Here, the thing to seal is each brief, approval and final version as you go: the client's sign-off on a piece of creative, the targeting settings sent to the platform, the AI tool and model version used to generate a video, licences for third-party material and the final cut as it was published. Sealing takes a fingerprint of the record, not the record itself, so nothing you need to keep confidential leaves your systems. If a transparency notice, a rights claim or a dispute turns on which version existed when, anyone you share the record with can check that it hasn't changed since it was sealed.

What it doesn't do is show that the ad was lawful, that a licence covered the use or that a disclosure was adequate. It shows that a record of what happened existed at a given time and hasn't been rewritten since. That matters when the regulator, a rights holder or an opposing candidate asks a pointed question.

To use it, your systems call WitniumChain through the REST API, the SDK or the MCP server at each step: when the brief is agreed, when a model generates a version, when the client approves and when the ad goes live. Each call returns a witness ID. Later, if the record is questioned, you calculate the fingerprint again and compare it with the sealed one. A match means nothing has changed. The first 2,000 witnesses a month are free, and you can read the documentation to see how it fits your workflow.

How it would work

1. Your campaign management system seals the brief, each AI generation event (tool, model, settings), approvals and the final ad as published.
2. Each seal returns a witness ID that you store alongside the record.
3. When a transparency request or a dispute arises, you share the record and the witness ID. The recipient checks the fingerprint and sees whether it matches.

What this means for you

If you run campaigns, place political ads or provide a platform for them, ask:

• Do you know which of your campaigns count as political advertising under the regulation's definition?
• Can you produce the prescribed records for any campaign from the past seven years, unchanged?
• If you use AI to generate creative, do your records show which tool, which model version and which settings were used?
• Can you show that targeting matched what the transparency notice said?
• If an approval or a licence is questioned, do you have the version that was approved or the terms that applied at the time?

The repository is live, the retention period is seven years and the fines are set as a percentage of revenue. Deciding how to keep a provable record is a question to answer now, not when someone asks for it.

What to watch

The Commission has scheduled an implementation dialogue on the regulation's application for 4 November 2026 ^[1]. Member states are setting up their own supervisory arrangements, and the first enforcement decisions will clarify how broadly the definition of political advertising is applied in practice.

Article 50 of the AI Act, which requires transparency when AI generates or manipulates content, applies from 2 August 2026 and overlaps with the political advertising rules ^[3]. A campaign using AI to generate video or text will need to meet both the political ad transparency requirements and the AI Act's disclosure duties, and keeping a checkable record of what was generated, when and by which system helps with both.

Sources

1. European Commission, "Implementation dialogue on the application of the Political Advertising Regulation with Commissioner Michael McGrath", event notice, 4 November 2026, https://commission.europa.eu/get-involved/events/implementation-dialogue-application-political-advertising-regulation-commissioner-michael-mcgrath-2026-11-04_en
2. Regulation (EU) 2024/900 on the transparency and targeting of political advertising, Official Journal of the European Union, 19 March 2024, https://eur-lex.europa.eu/eli/reg/2024/900/oj
3. Article 50, Regulation (EU) 2024/1689 (AI Act), AI Act Service Desk, https://ai-act-service-desk.ec.europa.eu/en/ai-act/article-50
4. Brussels Times, "New EU political ad repository targets transparency after regulation takes effect", https://www.brusselstimes.com/2357581/new-eu-political-ad-repository-targets-transparency-after-regulation-takes-effect