Skip to content

Legal

AI transparency

How C Wire uses AI, what it reads, what it never reads, and who decides. Written for clients, publishers and regulators alike.

01

Last updated: 11 September 2026

C Wire AG ("C Wire") uses artificial intelligence to plan, buy and report advertising. This statement explains, in plain language, where AI is used, what it processes, where people decide, and how we position ourselves under the EU Artificial Intelligence Act. It complements our Privacy Policy and applies to cwire.com, the C Wire Console, the Try MatchPersona demo and the campaigns we run for clients.

1. Where we use AI

BrandDNA and CampaignDNA. Language models read a brand's own material (its website, guidelines, existing communication) and a campaign brief, and derive two structured records: what the brand means, and what this campaign wants. Every field keeps its provenance (read, inferred or set by a person) and each record carries a quality score.

MatchPersona. From those two records, an agentic pipeline generates personas: groups of people described by life situation and by the topics where they spend attention, never by identity. It validates them against automated tests, then matches their topics to content: articles on the Trusted Web and, as a separate product, channels and videos on YouTube. Every channel or video on a YouTube plan carries the reason it was selected.

MatchOutcome. A bidding algorithm prices each impression against the one metric the client agreed to before the campaign, using the MatchPersona match and the Placement Graph (what each audited ad unit has delivered before). It optimises on delivered performance, not on a proprietary score.

Reading content. Before any bid, C Wire reads the specific page, video or app screen and records what it is about in the Content Graph. If content cannot be read, it is not bought.

2. What our AI processes, and what it never processes

It processes: publicly available editorial content (pages, videos, app screens), a brand's own published material and guidelines, campaign briefs and landing pages supplied by the client, and delivery data at the level of placements, personas and campaigns.

It never processes: personal data about readers or viewers. There are no cookies, device identifiers, audience files, behavioural profiles or cross-site tracking anywhere in the planning, targeting or bidding. Personas describe groups through content, not individuals. Reporting contains no user-level data. Technical identifiers used for ad delivery and frequency capping are described in the Privacy Policy.

3. Where people decide

Clients review and approve BrandDNA, CampaignDNA and the personas before anything runs, and can exclude any persona, topic, channel or video. The metric a campaign is judged on is agreed by people before launch. The Trusted Web is admitted, audited and removed by C Wire staff against published criteria. AI proposes; people approve, exclude and answer for the result.

4. Limits we state openly

C Wire is not a brand safety or verification vendor; clients keep their verification partner for measurement. Personas are inferences about where groups of people spend attention, not facts about individuals. Models can misread content; that is why content is read before the bid, why clients review before launch, and why anything that cannot be read is excluded. We do not claim results we have not measured.

5. The EU Artificial Intelligence Act

C Wire is a Swiss company whose outputs are used in the European Union, so the Act applies to us. Our current assessment, reviewed with counsel and revisited as guidance evolves:

Planning and targeting (BrandDNA, CampaignDNA, MatchPersona) are not high-risk systems under Annex III and do not interact with natural persons in a way that triggers Article 50 labelling. Advertising bidding (MatchOutcome) is not a listed high-risk use case; we treat the Article 5 prohibitions on manipulative or vulnerability-exploiting practices as a design constraint and maintain suppression rules so that sensitive-category advertising is not placed against sensitive contexts.

We do not currently offer AI-generated advertising creative as a public feature. Should we do so, we will mark such outputs in a machine-readable way as Article 50 requires.

Article 4 (AI literacy) applies to us as provider and deployer; our staff are trained on the systems they operate. We publish this statement and our data boundaries so that clients can meet their own transparency duties, including under the Digital Services Act.

6. Questions and contact

Questions about how AI is used in your campaign, requests to review or delete a BrandDNA, CampaignDNA or persona record, and questions from regulators: privacy@cwire.com, or C Wire AG, Grubenstrasse 54, 8045 Zürich, Switzerland.

Questions

AI transparency, explained

Does C Wire use my customers'' data to train or run its AI?

No. C Wire''s AI reads content and a brand''s own material. It does not process reader or viewer data, and client campaign data is not used to train models for other clients.

Is any of the advertising itself generated by AI?

Not as a public feature today. Creative is supplied by the client. If C Wire offers AI-generated creative in future, it will be marked in a machine-readable way as the EU AI Act requires, and this statement will be updated.

Can I see why a placement was chosen?

Yes. Personas are reviewed before launch, and on YouTube every channel and video on the plan carries the reason it was selected. On the Trusted Web reporting goes down to the audited ad unit.

Who is accountable for a campaign decision?

C Wire. AI proposes personas and bids; people at C Wire and at the client approve, exclude and answer for the result. Commercial terms name the accountable party.

Where is the processing done?

On infrastructure in the European Union and Switzerland. Details of sub-processors are available on request under our data processing agreement.