Privacy Policy

This policy covers two very different things: the information you give us as a customer (an account, a subscription), and the public records we publish from: property records, which can name real people as owners of record, buyers, sellers, permit applicants, or attorneys, and public meeting records. Both are explained below.

• Account: your email address, a display name you choose, and a hashed password. We never store the password itself.
• Billing: handled by Stripe. We store your Stripe customer id, your plan and its status, and the brand and last four digits of the card on file so the account page can show it; for one-time purchases, the product, the parcel it was bought for, the amount, and the Stripe session id. Full card numbers never reach our servers.
• Usage: standard server logs (IP address, browser, pages requested, timestamps) kept for security and debugging, plus your watchlist and saved preferences. We also collect anonymous page-family counts, research-action counts and a 20% sample of browser performance measurements. These measurements contain no account identifier, full URL, search text, parcel number or IP address; they use no persistent tracking identifier, honor browser Do Not Track and Global Privacy Control, and are deleted after 35 days. When browser Do Not Track and Global Privacy Control are not enabled, a tab stores only the landing-page category, whether a recognized search engine referred the visit, the device category and, when present, a campaign label from our fixed public list; we use this context for 30 minutes. Arbitrary campaign text and advertising click identifiers are discarded. We use an update-feed opening timestamp for up to 35 days to count later visits after that action. Expired values are replaced or removed on a later measured visit. Neither value is a unique browser identifier. At checkout, those coarse categories accompany the payment metadata sent to Stripe. Short-lived paid-start counts use an opaque event key to avoid counting the same confirmed checkout twice; the measurement table does not store the Stripe session id.
• Checkout attribution: for checkouts with permitted measurement, we also keep a durable server-side financial reporting record: the product, purchased town for town-specific products, checkout date, amount, currency, coarse entry and campaign categories, and later refunds or disputes. It uses hashed checkout and payment keys for reconciliation, with no account identifier, raw Stripe identifier, parcel number, URL, IP address or browser identifier in that record. This payment reporting history is retained beyond the 35-day browsing window so a late refund or repeated payment event can be reconciled correctly. Only authorized staff can view its grouped reports. It does not track subscription renewal revenue or visits across sites.
• Research updates: followed records, published snapshots, detected differences, read status and email preferences are stored with your account. Private notes in the existing watchlist remain in your browser. We retain detected updates and their email delivery records for up to 90 days; deleting your account removes its monitors and update history. Messages you request include a link to stop that record’s emails. Provider delivery events confirm mail-server delivery; we do not use open pixels to track whether you read a message.
• AI questions: when you use AI chat, your question, recent conversation context and the relevant records your plan can access are sent from our Vercel-hosted server to OpenAI to prepare an answer. Questions you submit may also be sent, without your account or IP address, to TypeSafe to route them to the right answer (a question containing a digit, a dollar sign or a percent sign is not sent); TypeSafe processes them under its own retention policy and does not use them to train models. This site does not keep questions in its database or application logs, except that a question in an /ask page address appears in our standard server request logs, which are kept for 30 days. We do not save conversation history. Providers process the requests under their own retention policies; do not include confidential information. We keep usage counts keyed by a one-way hash of your account id or IP address to enforce allowances. These counts expire within 40 days and are deleted during a later request. They contain no question text or raw IP address.
• Email: transactional messages only — receipts, password resets, service notices and record updates you explicitly enable. No marketing list unless you opt in later.

When you connect an assistant to our public research server, it sends the lookup inputs you select, such as the town, a property address or a parcel number. GroundSignal returns dated public parcel fields and source links. You can use this service without a GroundSignal account. We count requests to limit abuse. Your assistant provider handles your conversation under its own privacy policy.

We set a session cookie so you stay signed in, and a short-lived cookie that caches your session to speed up page loads. We do not run advertising or cross-site tracking cookies. We measure page views and a few product steps with Vercel Web Analytics, which works without cookies or a persistent identifier, honors Do Not Track and Global Privacy Control, and receives page addresses without query strings or parcel numbers. Stripe sets its own cookies during checkout under its own policy.

The property records on this site are obtained from public datasets published by the Cook County Assessor, the Cook County Board of Review, Cook County GIS, FEMA, and the Illinois Department of Revenue (MyDec PTAX-203). They include, for each parcel, the taxpayer (owner) of record and the mailing address on file with the Assessor, the names of buyers and sellers on recorded sales and MyDec declarations, the applicants on building permits, and the attorneys or agents on assessment appeals. MyDec buyer, seller, agent, and preparer names are treated as identity and are not shown on the free layer.

We also publish inferences drawn from those records — for example, that two owners share a mailing address, or that two people at the same address share a surname. Every inference is labeled as such in the product and is never presented as a verified fact.

We collect public meeting records — agendas, agenda packets and minutes — from the City of Park Ridge’s public meeting archive. We keep the original files privately; we do not republish them, and our pages link to the City’s official copy with the page number. We read these records to publish findings and records labeled AI-assisted research, and only after software checks every quote against its page (see the Methodology).

• Pages about private individuals are not extracted. Pages we identify as being about a private person, such as résumés, letters from residents, public comment and medical information, remain only inside the original file we keep privately: their text is not extracted, stored or searched.
• How the records are read. AI research tools read the other extracted pages to propose findings, and TypeSafe may screen what they propose. This research step sends public-record text only, never information about you as a visitor (your chat questions are covered under “AI questions” above). Nothing a model proposes is published unchecked: software checks every quote word for word against its page and computes every number and date, and the privacy and wording checks must pass.
• Private people in these records are never named, apart from the applicants and petitioners described in the next item. Where our research needs to refer to any other private person, it records a one-way key (a keyed hash of the name) instead of the name, so the same person can be recognized again without being identified.
• Who may be named. Elected officials appear in their public role, and City staff by title. Companies may be named, and so may applicants and petitioners, whether people or companies, who ask the City for a variance, a permit or a designation. Residents, people who spoke or wrote in public comment, individuals paid by the City and people appointed to boards are described only by role.
• Questions about these records. Chat questions are handled as described under “AI questions” above: they may be processed by OpenAI to prepare an answer and by TypeSafe to route them to the right answer.
• Withdrawal. If a published finding or record describes you, or you believe it should not be public, email support@cookcountyrecords.com with the page address. We aim to answer within ten business days. The owner of this site can unpublish any finding or record in one step; an unpublished item is removed from the site and from chat answers and is not published again as it was.

If a record about you is wrong at the source, the fix has to happen at the agency (for example, an incorrect taxpayer name is corrected with the Cook County Assessor); our copy updates on the next refresh. We will tell you which agency to contact.

If you are an owner of record and do not want your name displayed, email support@cookcountyrecords.com from an address that lets us confirm you own the parcel. Where the law permits, we will suppress the name on our site within ten business days. Lead lists downloaded after that date exclude the name, and we notify every buyer who downloaded a list containing it in the previous twelve months so they can delete the row; copies already downloaded cannot be recalled. We give particular priority to requests from people with a safety concern, such as those covered by an address-confidentiality program.

Customers can ask us to export or delete their account data at any time. Deleting an account removes your profile, watchlist, and our billing cache; Stripe retains payment records as required by financial regulations.

We do not sell the account information you give us. We do sell compilations of public property records — including the county’s owner of record for each parcel — to businesses as lead lists, under a license that forbids resale and consumer-report uses (see the Terms). If you are an owner of record and do not want your name in those lists, the suppression request above covers them too.

The services that process data on our behalf are:

• Stripe — payments and subscription billing.
• Vercel — hosting, delivery of the website, routing AI requests and private storage of the original meeting-record files. Its cookie-free web analytics is described under cookies above.
• OpenAI — preparing AI answers from your questions and relevant records.
• TypeSafe — routing the questions you submit to the right answer, and screening public records.
• Neon — the managed database that stores accounts, the published records and the text extracted from collected meeting records.
• Resend — delivery of transactional email.

We may also disclose information when required by law or to protect the rights, safety, or property of our users, the public, or ourselves.

Account data is kept while the account exists and for 90 days after deletion in backups. Server logs are kept for 30 days. Published record versions are retained for rollback and then pruned; a name suppressed at your request stays suppressed across future refreshes.

The service is not directed to children under 13 and we do not knowingly collect account information from them.

We will post changes here and email subscribers about material ones. Questions or requests: support@cookcountyrecords.com, or see the Contact page. Written requests: GroundSignal, 205 West Touhy Avenue, Suite 328, Park Ridge, IL 60068. Our methodology — including which fields are facts and which are estimates — is on the Methodology page.