Room to think.

Editorial suggestions

Story tips and requests to check claims go to a private editorial inbox, never directly onto the site. No name or email is required. Please leave out sensitive personal information. Suggestions expire after 90 days and are removed on the next inbox visit or submission. A daily rotating keyed hash of the network address limits spam; the address itself is not stored. Rate-limit records expire within 24 hours and are removed on the next submission.

What Bright collects

Bright has no public account system, advertising trackers, analytics cookies or third-party analytics service. The owner’s private audience dashboard uses ChatGPT sign-in. Search runs in your browser. Selecting a record updates the page address so you can return to it.

Traffic and site health

Bright counts page views, approximate tab visit starts, a first internal continuation in a tab, and a return-tab start inferred only from one local timestamp between 12 hours and 30 days. It also counts pages visible for at least 30 seconds, pages with an external-link click, coarse screen size, loading-performance categories and browser-error categories. Referrers and supported campaign links are reduced to fixed labels such as X, TikTok, Google or launch before they are sent. Unknown referrers become “other”; search terms, full referrer URLs, query strings, error messages and stack traces are not collected.

These measurements go only to Bright’s own server and are added into hourly totals by public page and category. No visitor identifier, IP address, user agent, fingerprint or individual browsing history is stored by this measurement code. Aggregate totals expire after 400 UTC days. They do not identify unique people or returning users.

Visit starts are also added into country totals and, for U.S. visits only, two-letter state totals supplied by the hosting edge. Bright does not store IP addresses, cities, precise locations or a visitor-level location history. The hosting provider derives these coarse locations at its network edge. Bright’s application does not read the underlying raw IP for analytics; provider access logs and their retention are separate. No earlier locations are inferred.

This tab’s session storage holds a last-active time and coarse source and campaign labels to avoid counting each page as a new visit. It expires after 30 minutes of inactivity or when the tab closes and is never sent as an identifier. A separate local return timestamp is replaced on each new tab visit and is not sent. If storage is unavailable, a page load may count as a new visit. Global Privacy Control and Do Not Track stop both traffic and experience measurement. Owner-authenticated traffic and browsers the owner explicitly excludes are not counted.

What the scale experience counts

To learn whether the scale, Now and the Index work for people, the pages send a few anonymous counts to Bright’s own server: which scale was opened, reached or tried (for example “played the chirp” or “moved the field divider”), whether you jumped or scrolled there, which record you opened from a scale, which clock you chose in Now and how you sorted the Index. Each count carries only a coarse context — narrow or wide screen, touch or pointer, reduced motion on or off.

The counts contain no cookie, identifier, device fingerprint, page address or free text, and Bright’s code does not read or keep your IP address, user agent or referrer with them. Counts are added into daily totals, never kept as a history of one visit. If your browser sends Global Privacy Control or Do Not Track, nothing is counted at all. Business interaction counts use a separate daily, event-only aggregate. They contain no form values or reader history.

Public API operations

Bright writes a small structured event to its hosting logs when a public discovery, guide, model, Bright Weekly or claim-evidence API responds. It contains the API operation, a fixed topic category, status, result count, empty-result state, response time and contract version. It does not contain the question, location, device or platform text a caller supplied, an IP address, user agent, user identifier or request identifier. Bright uses exported aggregate logs to check latency and error rates; it does not maintain a visitor-level API history.

The assistant connector at /api/mcp works the same way. For each request it logs the protocol method, the Bright tool called, whether it succeeded, response time and contract version. When an assistant first connects, Bright also records a coarse assistant family (for example Meta, OpenAI or Anthropic) derived from the client name the assistant sends; the name itself is not kept. It does not log the question, search text or claim an assistant sends, or any account, user or device identifier. The connector needs no sign-in. It receives the tool requests your assistant sends, including their question or search terms, but cannot browse your other conversations or files.

Hosting and external sources

Hosting providers may process technical request data such as IP address, browser information, requested URL, and errors to deliver and secure the site. Their infrastructure logging is separate from Bright’s application behavior.

Original-source links take you to independent websites with their own privacy practices. No third-party video or audio loads automatically.

Reading preferences on this browser

Bright can remember followed questions, read records, a continue-reading link, prior catalog checkpoints, your preferred Scale or Latest view, and newsletter invitation choices in bounded first-party browser storage. These preferences stay on this browser and device. Shared-browser users share them. They are not sent with analytics or synchronized with an email address. The reading controls let you reset reading history and followed topics without clearing unrelated browser data.

Bright Weekly

When email signup is available, Bright uses Beehiiv to manage newsletter addresses, affirmative consent wording and version, signup source and timestamp, confirmation, delivery and unsubscribe preferences. Signups use the publication’s opt-in setting; an existing unsubscribe is not silently reversed. Provider-hosted links handle preferences and unsubscribe. Beehiiv is the subscriber system of record.

Bright’s server safety ledger uses keyed pseudonymous address digests for duplicate prevention and suppression. It does not keep a second email list. Suppression records can outlast active subscriptions to prevent unwanted contact; deleting an address from the active list does not mean removing its do-not-contact protection. Contact hello@brightaifuture.com for access or deletion requests. Provider processing may take place outside your country under its applicable terms.

When collection is enabled, short-lived keyed network digests limit automated requests. Raw network addresses are not stored in this ledger or routine application logs. These anti-abuse controls are separate from optional analytics; Global Privacy Control does not disable the security needed to process a form you submit.

Contact and partnership inquiries

Current contact links open your own email application. They do not subscribe you to Bright Weekly. The optional web inquiry form remains disabled until a verified delivery service and retention process are configured. When enabled, it sends your name, work email, organization, interest, timing, optional budget and message through Resend to Bright’s contact inbox. No message or address is placed in analytics, a public URL, or a local inquiry database.

The proposed inquiry retention policy is to remove unneeded inquiries within 90 days after a conversation closes, except records required for an ongoing agreement or legal obligation. The form cannot be enabled until the operator approves and implements that inbox/provider process. Avoid sending sensitive personal information.

Support payments

Payments are not enabled yet. If activated, Stripe-hosted payment pages process one-time or monthly support. Bright does not store card information or a local payment/customer database. Stripe and the approved account operator maintain financial records under their applicable retention obligations. Support never enrolls you in the newsletter.

Responsibility and questions

Brandon Brooks is Bright’s founder and editor. Bright AI Future operates Bright. For privacy questions, access or deletion requests, contact hello@brightaifuture.com. Third-party source sites, TikTok, newsletter and payment providers have their own privacy practices; ordinary outbound links do not load their tracking scripts on Bright.