Legal

Privacy Policy

Plain English. Effective July 2026. GDPR standards apply to every visitor regardless of jurisdiction.

What OpenWatch is

OpenWatch is a geopolitical and macro intelligence platform built on open-source news. It aggregates publicly available signals from news feeds, government filings, and market data to produce country threat scores, scenario forecasts, and sector analysis.

OpenWatch does not profile end users on the core platform. There is currently no paid tier, no advertising, and no behavioural tracking of your use of the core site. The one exception is optional companion games (see Companion games below), which track your in-game activity — streaks, badges, leaderboard standing — to run the game itself. The primary privacy consideration for this platform is not user account data but personal data of individuals mentioned in source news articles— explained in the next section.

News-derived personal data

When OpenWatch ingests a news article, an automated pipeline extracts named entities from the text: people, organisations, and locations. The names of individuals mentioned in published articles are stored alongside the article record in our database as part of signal scoring and display.

This data comes exclusively from already-published, publicly available sources. OpenWatch does not independently collect personal data about individuals, does not profile private individuals, and does not produce per-person dossiers. Named entities appear because they were referenced in published reporting.

News content routinely covers sensitive topics — political opinions, religious affiliation, health conditions, criminal proceedings. OpenWatch does not extract or store these as separate columns; they are present only within the underlying article text.

Retention:raw article records including named-entity data are kept on a tiered schedule set by each item’s significance score (0–10): routine items (scoring below 4.5 — the large majority) are automatically deleted 3 months after collection; moderately significant items (4.5–6.5) after 12 months; and the most significant items (6.5 and above — major geopolitical events) after 36 months. No raw article record is kept longer than 36 months. Derived scores, forecasts, and snapshots are retained longer as aggregated analytical output and do not contain raw entity lists.

Lawful basis:we believe our processing falls under legitimate interest (GDPR Art. 6(1)(f)) and the journalism and research exemptions that EU and UK member states apply to open-source intelligence work. We are reviewing this question with qualified data-protection counsel and will update this notice when that review concludes.

Prediction-market trader leaderboard

OpenWatch publishes an optional aggregate leaderboard that ranks public prediction-market traders by their on-platform track record. This data is drawn exclusively from the public APIs of Polymarket, Kalshi, and Manifold, covers accounts that already trade publicly under a self-selected handle or wallet, and consists of the handle, the public profile link, aggregate performance statistics (win rate, average position size, profit, a derived score and rank), and the public positions those platforms already display.

This is third-party public market data, notdata about your OpenWatch account. We do not link these handles to real-world identities, do not build per-person dossiers, do not monitor any individual’s activity, and do not send alerts when a specific trader acts. The leaderboard is an aggregate ranking of public performance only.

Lawful basis:legitimate interest (GDPR Art. 6(1)(f)). The trade activity is public by each platform’s own design; we ingest only what those platforms already publish and apply no special-category (Art. 9) processing.

Retention: leaderboard data is refreshed hourly from the platform APIs and retained only while the leaderboard is active and the source account remains public.

User account data

If you create an account or subscribe to alerts, we collect:

  • Email address— used to authenticate you and deliver the things you asked for. Nothing else.
  • Portfolio data— tickers and positions you add, stored server-side only when you are signed in. In local mode, portfolio data stays in your browser and never leaves your device.

We do not collect names, phone numbers, addresses, payment details, or any personally identifying information beyond email.

Companion games

When you play an OpenWatch companion game — currently Crowd Prophet on Reddit, via Reddit’s Devvit platform — we receive a platform-provided player identifier (for Reddit: your Reddit user ID and username) and store your activity within that game: fake-currency wager allocations, portfolio and payout history, badges earned, streaks, and leaderboard standing. No real money is collected, wagered, or paid out in any companion game. This is engagement data about your voluntary use of an optional game feature, not profiling of your use of the core OpenWatch platform.

Some of what a companion game displays is not literal, real-time human data. For example, Crowd Prophet’s “your sub vs the crowd” split starts from an auto-generated baseline (styled after that market’s real odds) on days when few real players have voted yet, with real votes layered on top as they come in — so the split is never a hollow 0/100, but isn’t 100% real users either. This baseline is disclosed here because it affects what “the crowd” showed you on a given day, not because it changes any game outcome.

Likewise, when live market data is temporarily unavailable or too thin/one-sided to fill a balanced board, some questions are drawn from a fallback set instead of Polymarket. Fallback questions resolve on their own declared, fixed schedule rather than a real-world event.

Companion games use publicly available prediction-market data (for example, from Polymarket) to generate daily questions. This is read-only reference content — it does not involve real trading through OpenWatch or the game, on any platform.

Linking to your OpenWatch account (optional).By default, your companion-game activity stays associated only with your account on that platform (e.g., your Reddit account) and is not connected to any OpenWatch.io account. If you choose to connect them, we display your platform username and the game data listed above on your OpenWatch profile. You can disconnect at any time — from your OpenWatch account settings once that control is available there, or in the meantime by emailing privacy@openwatch.io — which removes the game data from your OpenWatch profile without affecting your underlying game history on the host platform.

Retention:each companion game sets its own, generally short retention windows for in-game activity — for example, Crowd Prophet’s play-by-play records expire automatically 70–90 days after each round, cached usernames refresh every ~7 days, and score history is capped at your most recent 180 played days. Most companion games also offer a self-service in-game control — for example, Crowd Prophet’s “erase my data” — that removes your profile, streak, badges, and leaderboard standing immediately; day-specific play records not yet covered by that immediate removal age out automatically under the windows above. See that game’s own Terms & Privacy page for its exact figures.

Lawful basis: legitimate interest for unlinked gameplay (operating the game); consent for the optional OpenWatch account link. Covered by the same rights described in Your rights above (access, correction, export, restriction, objection, deletion).

Cookies

  • Strictly necessary— Supabase authentication session cookies keep you logged in. These are required for authenticated features and set without consent, as permitted for essential cookies.
  • Analytics— we use Vercel Analytics and PostHog for aggregate page-view metrics (page views, referrers, country-level geography). No cross-site tracking, no personal profiles, no selling of data. Data stored in the US. You can decline analytics via the cookie preference banner.
  • Preference storage localStorage entries for your cookie choice, theme, and dismissed banners.

We use no advertising, tracking, or third-party marketing cookies, no retargeting pixels, and no fingerprinting.

How to manage: use the cookie preference banner shown on your first visit, or your browser settings to clear or block cookies at any time.

Terms & Privacy agreement records

When you agree to our Terms of Use and Privacy Policy — at sign-in, or again if we make a change significant enough to affect your rights or how we handle your data — we keep a permanent record: your account, which document, which version, how you agreed (email link or Google sign-in), a timestamp, a hashed version of your IP address, and your browser’s user-agent string. This is the same function a signature and date serve on a paper contract.

These records cannot be edited or deleted once written, including after you delete your account — a deliberate exception described in the Terms of Use’s Data retention and legal holds section.

Each version of these documents has its own effective date and a short plain-language summary, shown to you before you agree. If we make a change significant enough to affect your rights or data handling, existing users are shown what changed and asked to agree again, or may decline and request account deletion instead, as described in Your rights.

Retention: kept indefinitely as proof of consent.

Lawful basis:contractual necessity (GDPR Art. 6(1)(b)) — recording your agreement is necessary to establish and administer the contract these documents form.

Sub-processors

We do not sell data. To run the platform we share data with the following infrastructure vendors, each acting on our behalf under a Data Processing Agreement or equivalent:

  • Supabase(US) — database and authentication. Stores account records and all pipeline data.
  • Vercel(US) — web hosting and edge analytics.
  • PostHog(US) — web analytics. Collects aggregate page-view data (page URL, referrer, country-level geography). No cross-site tracking, no selling of data. Data processor: PostHog Inc. (posthog.com).
  • Railway(US) — backend compute for the data pipeline, including NER enrichment and translation.
  • Anthropic(US) — LLM processing for enrichment, article summarisation, and translation fallback after DeepL. Full article text is sent during enrichment.
  • Google (Cloud Translation API)(US) — optional first-pass translation when a Google Translate API key is configured. Full article text is sent when active. Covered by Google Cloud’s Data Processing Addendum.
  • DeepL(EU / DE) — primary article translation. Full article text is sent. EU-based; no international transfer.
  • Voyage AI(US) — text embeddings for semantic search. Article text is sent for embedding computation.
  • Mapbox(US) — geocoding and globe map rendering. Place names (not personal addresses) are sent for geocoding.
  • Resend(US) — transactional and digest email delivery. Email addresses are shared when you have subscribed to alerts or digests.

All US-hosted processors handle EU/UK data under the EU-US Data Privacy Framework and/or Standard Contractual Clauses. We do not share data with advertisers, data brokers, or parties that would resell or repurpose it.

Retention

  • News signal records(including named-entity data): tiered by significance score — routine items automatically deleted 3 months after collection, moderately significant items after 12 months, and the most significant items after 36 months. Never longer than 36 months.
  • Watch briefs: automatically deleted after 48 hours.
  • Account and email records: kept while your account is active; reviewed and processed upon a verified deletion request.
  • Companion-game data: see the Companion games section above for each game’s specific retention windows and self-service deletion controls.
  • Anonymous interaction data (signal bookmarks, wager votes): retained in de-identified form after account deletion; not linkable to your identity. You can instead permanently delete your activity history from the Account page at any time (see Your rights).
  • API cost logs: automatically deleted 12 months after recording.
  • Server and request logs: rotated within 30 days per Vercel and Railway defaults.

Retention exceptions. We may retain personal data beyond these periods where required or permitted by law, including: to comply with a legal obligation or regulatory requirement; for the establishment, exercise, or defence of legal claims; to investigate or document violations of our Terms of Use (including abuse or harassment); during any active legal hold; or for fraud and security incident investigation. Where data is retained under an exception, it is held securely and used only for the purpose of the exception.

Email opt-outs are processed immediately and unconditionally. Opting out of email does not constitute a deletion request and does not affect other stored data.

Your rights

Under GDPR, UK GDPR, CCPA, and similar regimes, you have rights to access, correct, export, restrict, object to, or delete the data we hold about you. We aim to honour those rights for everyone regardless of jurisdiction. Terms/Privacy agreement records are available via access or export requests but, per the retention exception above, are not subject to deletion.

Account holders can delete their account and manage email preferences from the Account page. You can also submit a formal privacy request below.

Deleting your activity history.Signed-in users can permanently delete their activity history — votes, signal bookmarks and interactions, watchlists and saved lists, and alert history — from the Account page at any time, separately from account deletion. Be aware that if information connected to you was publicly visible — for example a public portfolio profile or leaderboard entry on OpenWatch, or your activity in a companion game on Reddit — search engines such as Google, web archives, and other scraping services may retain their own copies. Deletion removes the data from OpenWatch’s systems; it cannot remove copies already held by third parties, which must be requested from those services directly.

Individuals named in news articles:rights to erasure under Art. 17 are subject to qualifications that apply to journalism, research, and legitimate-interest processing of publicly published information. The scope of those qualifications is part of our ongoing counsel review. Contact us with your request and we will review it.

We will respond to verified requests in accordance with applicable law. We may request an extension of the review period where permitted, and will notify you if we do so. Some requests may be subject to retention exceptions described above.

Submit a privacy request →

Security

All traffic is served over HTTPS. Authentication uses industry-standard token-based sessions. We apply automated data retention policies to minimise the amount of personal data held at any given time.

Changes to this policy

If we materially change how we handle data, we will update the effective date above and notify subscribed users by email. The latest version always lives at /privacy.

Contact

Privacy questions, access or deletion requests, or concerns: privacy@openwatch.io.

Effective: July 2026 — lawful basis for news-derived data under counsel review

3485b22