Your privacy and controls

Stored in the EU, on our own servers

Our core stack (the app, the databases, object storage and search) runs on servers we manage in Germany, rented from Hetzner. No cloud giant hosts it for us.
Your account, reading history and personalisation are stored there, under European rules. Some features need a provider outside the EU, such as Pollar Chat, search, Radar audio, sign-in emails and payments. Our privacy policy lists every one of them.
Self-hosting our core keeps your account and reading data on infrastructure we control. Cloudflare, a US company, delivers the site to you and protects it from attacks.

We measure traffic with Umami, which we self-host on our own infrastructure, never a third-party cloud.
It is cookieless: no cookies, no localStorage, no cross-site identifier. It groups visits with an identifier that changes every month, so it cannot follow you over a longer period or across sites.
It records page views, referrer, country and browser details, and never your IP address. Processed under legitimate interest (Art. 6(1)(f) GDPR), and switched off when your browser sends Global Privacy Control or Do Not Track.

Like any modern publisher, we run advertising and measure whether it works. We use only Meta, and only for conversions: joining the waiting list and buying a membership. We send hashed identifiers, never your raw email, never page views and never the articles you read.
In the European Economic Area, the United Kingdom and Switzerland this runs only if you consent. You can turn it off anytime in the consent banner or in Settings, then Privacy, and we honour the Global Privacy Control signal.
This measurement data goes to the United States, where Meta is certified under the EU-US Data Privacy Framework, with standard contractual clauses as a fallback.
Everything else holds: we don’t sell your data, we don’t profile what you read for advertisers, and your account and reading history stay on our servers in the EU.

Pollar’s texts are written by AI, and the newsroom corrects errors after publication. Every AI text carries a visible label: in the feed, in the article, in the brief and in audio. We never pass off machine-written text as written by a person.
Our sourcing, methodology and corrections process are published openly in our editorial standards, aligned with European media-credibility norms (EMFA).
Published in four EU languages: English, Polish, German and French. Each member state is treated as equally relevant.
The AI that reads what you type for your feed, your Radar topics and members’ posts runs in the EU, through an EU gateway to Google, Microsoft and AWS data centres. We don't send your email or account identifiers with it. Pollar Chat and search are processed in the US; our AI page maps every feature.

Pollar is built to WCAG 2.1 level AA, the standard the European Accessibility Act requires: full keyboard operation, screen-reader text alternatives (even for data charts), and a 4.5:1 contrast ratio in light and dark. Colour is never the only signal.
Accessibility is ongoing engineering work, not a launch-day checkbox. We audit it as we ship every change.
We publish a full accessibility statement with a direct channel to report any barrier.