Privacy Policy

Our Philosophy

Dreamfields exists to serve dreamers, not to harvest data. Your inner life is deeply personal, and this platform is built around respecting that. Privacy is not a feature we bolt on — it is foundational to how Dreamfields works. We collect only what we need to run the platform, we are transparent about what we do with it, and we give you full control over your own data.

Your Data Belongs to You

Everything you create on Dreamfields — your dream entries, comments, media, journal notes — is yours. You own it. We don't claim any ownership over your content. You can export your data or delete it entirely at any time. When you delete something, it is actually deleted, not hidden or archived on our end.

What We Collect

We collect the minimum needed to provide the service:

• Account information: Your email address, username, and display name — needed to identify your account
• Content you create: Dreams, comments, reactions, and media you choose to post
• Basic usage data: Error logs and minimal analytics to keep the platform running and fix problems
• Authentication tokens: Stored locally in your browser to keep you logged in

We do not use tracking pixels, fingerprinting, or third-party analytics services. We do not build advertising profiles. We do not track you across the web.

What We Will Never Do

• Sell your data to anyone, for any reason, ever
• Share your information with data brokers or advertisers
• Use your content to train AI models, or allow any provider to do so
• Show you advertisements
• Monetize your personal information
• Allow any third-party processor to retain, train on, or repurpose your content

Any third-party service we use for processing (such as our AI providers, Anthropic and Groq) is contractually prohibited from using your content for model training or repurposing it beyond service delivery. Temporary retention for system reliability may occur (see the AI Features and Third-Party Processing section below for provider-specific details). We only work with providers whose terms protect your data, and we review compliance regularly.

Visibility and Consent

Dreams you mark as private are visible only to you. Dreams marked public or shared with a circle are visible to the audience you choose. You can change the visibility of any dream at any time. Nothing is shared without your explicit choice.

Collective Dreamscape

Dreamfields analyzes dream content across the community to surface collective patterns — themes, symbols, and trends that emerge when many people dream about similar things. This is the Collective Dreamscape.

Public dreams are included in collective analysis by default — they are public. Circle dreams are analyzed only within their circle, never in the global collective. Private dreams are never included unless you explicitly opt in via the "Contribute to Collective Dreamscape" setting in your Radar settings.

Collective analysis is performed by sending dream content to our AI provider, Anthropic, via their API. No usernames, display names, email addresses, or account identifiers are included in these requests — only dream text. For aggregated trend readings (e.g., "what themes are emerging this week"), only signal counts are sent, not dream text at all. However, dream text may naturally contain personal details you wrote (names, places, etc.), and those are included as part of the text when individual dreams are processed for concept extraction. Anthropic processes this content in real time and does not retain it afterward. Anthropic's API terms prohibit them from using any data sent through the API for model training or any other purpose.

Your identity is never revealed in collective results — only aggregate patterns are visible (e.g., "48 people dreamed of frogs," not who those people are). If you opt out, your private dream contributions are flagged for removal and actively purged on the next pipeline cycle, which runs approximately every 6 hours.

Dream Mentions and Invites

You can tag people who appeared in your dreams. If you tag a registered user in a public dream, they will be notified. If you tag them in a private dream, you choose whether they are notified. If you enter an email address for someone not on the platform, we will send them a single invite on your behalf — we will never contact them again unless they sign up. We do not store email addresses of non-users beyond what is needed to process the invite.

AI Features and Third-Party Processing

Dreamfields uses two third-party AI providers: Anthropic (the company behind Claude) for text analysis, and Groq for voice transcription. Here is how your content is processed:

Automatic analysis on every dream. When you post a dream, Dreamfields automatically sends your dream text to Anthropic to detect dream signals (recurring themes like "flying," "water," "loss") and extract key concepts. This happens on every dream you post, regardless of visibility. No usernames, email addresses, or account identifiers are included in these requests — only the dream text itself (up to 2,000 characters for concept extraction, up to 5,000 characters for signal detection). However, dream text may naturally contain personal details you wrote (names of people, places, etc.), and those are sent as part of the text. This processing powers the signal tags and concept links you see on your dreams.

User-initiated AI features (dream interpretation, visualization) are opt-in — they only activate when you request them. When you use one of these features, the content of the specific dream you selected is sent to Anthropic for processing.

Collective Dreamscape analysis is a separate, periodic process where Dreamfields sends batches of dream content to Anthropic to identify community-wide patterns, trends, and emerging themes. This processing happens automatically for eligible dreams (see the Collective Dreamscape section above for eligibility and opt-out rules). No usernames, email addresses, or account identifiers are included — only dream text. For aggregated trend readings, only signal counts are sent, not dream text at all.

Voice transcription. If you use the voice recording feature to dictate a dream, your audio is sent to Groq for transcription using their Whisper speech recognition API. The audio is processed in real time and converted to text. Groq may retain your audio for up to 30 days for system reliability and abuse monitoring, after which it is deleted. Groq does not use your audio to train AI models. Groq's infrastructure runs on Google Cloud Platform (GCP) in the United States, meaning your audio is processed on US-based servers. The resulting text is stored on Dreamfields as your dream entry, and the original audio file is stored on our servers. Voice recording is entirely opt-in — it only happens when you choose to record.

Anthropic processes content in real time via their API and does not retain it afterward. Under Anthropic's API terms, data is not used for model training, not stored beyond the processing request, and not shared with any third party.

Groq processes content in real time but may retain inputs for up to 30 days for system reliability and abuse monitoring. Groq's Services Agreement explicitly prohibits using your data for model training. Groq uses infrastructure sub-processors (such as Google Cloud Platform) to provide the service, but does not share your data with third parties for any other purpose.

Neither provider is permitted to use your content for training AI models. We review our providers' terms regularly and will not use any provider whose terms do not protect your data.

Anonymous Journals

If you start an interpretation without signing in, Dreamfields creates an anonymous account tied to a long-lived browser cookie (df_anon_uid) so your journal survives across reloads and new dreams. You do not need to provide an email address.

Retention. Your anonymous journal is kept until one of the following happens:

• You delete it through the in-app control (immediate soft-delete plus a 24-hour grace window in case you change your mind, then permanent purge).
• You clear cookies and never return — in which case the empty scaffold is purged after 30 days if and only if it contains no dream content. Scaffolds that hold any dreams are retained until you delete them or attach the journal to a real account.
• You attach the anonymous journal to a Google or email account, after which the standard signed-in retention rules apply and the journal becomes a normal Dreamfields account.

We do not sell your dream content. We do not train external models on your dream content. The same protections that apply to signed-in accounts apply to anonymous journals.

AI provider. Anonymous interpretations are generated by sending the dream text and a small amount of lens metadata (e.g. which interpretation lens you chose) to our AI provider, Anthropic. Anthropic retains API requests for up to 30 days for abuse monitoring per their Trust Center, after which the requests are deleted. We do not opt into training-data sharing, and Anthropic's API terms prohibit them from using API content for model training. Source: privacy.anthropic.com.

Pre-existing anonymous journals (retroactive disclosure). Anonymous journals captured before this policy section was added (April 26, 2026) are subject to the same retention contract described here — perpetual unless deleted. No earlier policy quietly promised a shorter window; this section simply makes the perpetual-retention contract explicit so future you can find it.

Analytics & Error Monitoring

Analytics are off by default. Dreamfields uses minimal anonymous analytics (funnel events such as "page viewed", "paste submitted") only when you explicitly opt in via the consent banner. The banner is non-blocking; declining (or ignoring) it leaves analytics disabled and does not affect your ability to interpret dreams, save journals, or use any other feature.

Error monitoring. When the platform encounters an unexpected error we send a diagnostic event to Sentry. Anonymous-interpretation routes (/api/v1/interpret/anon*) are wrapped in a redaction filter that strips dream text, query strings, cookies, and authorization headers before any event leaves the server. Recursively-nested fields (e.g. extra context, breadcrumbs) are scrubbed in the same pass. If the redaction logic itself ever raises an exception we drop the event entirely (fail-closed) rather than risk leaking content.

Pre-redaction window (retroactive scrub). Error-monitoring events captured before April 26, 2026 may contain anonymous dream content recorded as part of error context before the redaction filter was installed. On the day this filter shipped, we ran a one-shot scrub script against our error-monitoring provider to bulk-delete events matching the affected anonymous-interpretation routes. Any residual events that escaped that scrub expire from Sentry's default retention window within 90 days of capture, so by July 25, 2026 no anonymous dream content remains in our error logs.

Where Your Data Lives

Your data is stored on a managed PostgreSQL database hosted by DigitalOcean in Toronto, Canada. Media uploads are stored on encrypted block storage in the same region. Connections to the database are encrypted. When content is sent to our AI providers for processing (as described above), it is transmitted to Anthropic's and Groq's servers in the United States. This processing is transient — your data at rest lives in Canada. We do not replicate your data to third-party services beyond what is described in this policy.

Security

We take reasonable measures to protect your data — encrypted connections, secure authentication, rate limiting, and regular security reviews. However, no system connected to the internet is perfectly secure. In the event of a data breach, we will notify affected users promptly and transparently, disclose what was accessed, and explain what steps we are taking. We are honest about this: we do our best, but we cannot guarantee that a breach will never happen.

Data Deletion and Portability

You can delete your account and all associated data at any time from your account settings. Deletion is permanent — we do not keep shadow copies. We respect your right to leave and to take your data with you. If you need a full export of your data, contact us and we will provide it.

Cookies

Dreamfields uses only essential cookies and local storage for authentication and preferences. We do not use tracking cookies, third-party cookies, or any cookie-based analytics. There is no cookie banner because there is nothing to consent to beyond basic functionality.

Changes to This Policy

If we make meaningful changes to this policy, we will notify users directly through the platform — not buried in a changelog. Minor wording clarifications may happen without notice, but the spirit of this policy will not change: your data is yours, we don't sell it, and we respect your privacy.

Contact

Questions, concerns, or data requests? Reach out at dreamfields@phong.com. We read every message.