DFOS Data and Privacy FAQs
JUNE 4TH, 2026
We’re building DFOS as infrastructure for shared private internets. Here's what that actually means for your data in plain language.
The Short Version
When it comes to data and privacy, DFOS is a locked room, not a safe. Private from other users and the public internet, not private from the platform operator or infrastructure providers.
The portability piece is what's genuinely different. When you export your content from DFOS, it comes with a built-in timestamp and proof of authorship that no one can alter. Two years from now, on a different platform, you can still prove you made something here in 2026. Our current build (the SDK, public API, View Source) is what makes that promise real and independently verifiable.
The Longer Version
Ownership questions
Q: Do users own their content?
A: Yes. Users retain copyright over everything they post. DFOS asks for a license to host and display the content -- that's what lets the service work -- but that license doesn't transfer ownership. DFOS cannot claim they wrote it, sell it, or use it as their own. The legal term is a non-exclusive license. Think of it like giving a gallery permission to hang your painting -- they can display it, but you still own the painting.
Q: Do users own their identity?
A: Yes. Every DFOS account is built on a Decentralized Identifier (DID) -- a cryptographic identity that belongs to the user, not to DFOS. DFOS cannot revoke, reassign, or transfer your DID. This is a genuine differentiator from most platforms, where your username and history live entirely in a company's database and disappear if the company does.
Q: If a user posts something and then leaves a space, who owns that post?
A: The user still owns the copyright. However there are now multiple parties holding a copy:
- The user can export their post at any time with, original timestamps and proof of authorship intact.
- The space keeps the post visible. The Space Administrator decides whether it stays, same as messages stay in a workspace after someone leaves.
- DFOS retains it on their servers as part of operating the platform.
Owning the copyright and owning a copy are different things. The user owns the copyright permanently. DFOS and the space hold copies with a right to display them.
Q: What does it mean that content is portable on DFOS?
A: This is the part that's genuinely different from most platforms.
A writer, artist, or musician posts work on DFOS and exports it. Ten years later DFOS is gone, , or the platform evolved. The creator uploads that export to a new platform, a personal site, anywhere -- and proof of when they created it travels with the content. Not just "trust me, I made this in 2026" -- an independently verifiable timestamp that no one can alter, attached to the content itself. Useful for establishing creative priority, copyright disputes, or just having a portable record of your work that doesn't depend on any platform staying alive.
On most platforms, proof that you wrote something only exists inside that platform's database. If the platform disappears, the proof disappears too. On DFOS, the proof is baked into the content itself. Anyone with the right tools can verify you wrote it and when, without asking DFOS anything. (Think "browser view-source meets Etherscan.")
Privacy and encryption questions
Q: Is DFOS a private platform? What does "private" actually mean here?
A: DFOS is private in a specific sense: content inside spaces is access-controlled. Only members of a space can see what's posted there. It's not broadcast to the public internet, not indexed by search engines, not visible to non-members.
What "private" does not mean here: the content is not end-to-end encrypted the way Signal messages are. DFOS as the platform operator can access content on its servers. AWS, which hosts the infrastructure, has physical access to those servers.
DFOS is a locked room, not a safe. Access is controlled and restricted, but the platform itself holds the key, not just the user.
Q: Is content encrypted?
A: Partially -- and it matters to be specific.
- Cryptographic signing keys (the private keys behind each DID) are encrypted at rest using AWS KMS, hardware-backed encryption. They are only decrypted momentarily when a user posts something, then cleared from memory.
- Post content, messages, and other user-generated content is stored on AWS servers, protected by access controls (only space members can read space content) but not end-to-end encrypted. DFOS as the operator has access.
This is standard for cloud-hosted community platforms -- end-to-end encrypting content at the post level would prevent moderation, search, and AI features inside spaces.
Q: Who can actually see what's inside a private space?
A: In practice: space members, Space Administrators, and DFOS as the platform operator. The following can also access data in limited, defined ways:
- AWS (as the infrastructure provider, has physical access to servers).
- Anthropic (when a user invokes an AI feature inside a space, the relevant content is sent to Anthropic's API to generate a response).
- Stream (provides real-time chat infrastructure, so chat messages pass through their systems).
- Other service providers (each one sees only what their specific service requires).
"Private" means private from other users and the public internet, not from all infrastructure providers.
Q: Does DFOS sell user data?
A: No. DFOS does not and will not sell or process personal information for interest-based advertising. Revenue comes from platform subscriptions and transaction fees only.
Q: Does DFOS train AI on posts from private spaces?
A: No. DFOS does not use the content of private spaces to train machine learning or AI models. When a user invokes an AI feature, their prompt is sent to Anthropic as a live API call to generate a response -- that is not the same as training data collection. Anthropic's Commercial Terms of Service prohibit use of API data for model training, so that protection applies on both ends.
Infrastructure questions
Q: Who owns the servers DFOS runs on?
A: Amazon Web Services (AWS). DFOS rents cloud infrastructure from AWS, standard practice for virtually every platform at this stage. DFOS controls access to the data stored on it; AWS owns the physical hardware.
Q: Where is data stored geographically?
A: DFOS is a US-based company (Metalabel, Inc., New York). Data is stored in the United States. EU/UK data protection laws may offer stronger protections than US law.
User rights questions
Q: Can users export and take their data?
A: Yes. Users can request a full export at any time by contacting info@dfos.com. The export includes:
- profile information
- all content they created (with original timestamps and proof of authorship intact)
- a list of space memberships
- transaction history where applicable.
The proof of when you created something travels with the export -- you are not dependent on DFOS remaining operational for that proof to hold.
Q: Can users delete their data?
A: Yes, with one exception. Users can request deletion of their account and personal information. DFOS removes the account and disassociates personal information from active services.
The exception: cryptographic proof records (DID, public keys, content signatures) are retained permanently. These records are pure cryptographic commitments -- hashes of hashes -- not personal information in any meaningful sense. They are the mathematical mechanism by which authorship can be verified independently of DFOS, and cannot be deleted without breaking the integrity of the proof chain.
Q: What happens to posts if a user deletes their account?
A: Posts remain in spaces where they were posted. The Space Administrator decides whether they stay visible. Content remains associated with the user's DID -- authorship is still attributed -- but the active account is deleted. Same as how most platforms handle account deletion.
Q: What happens to user data if DFOS shuts down?
A: DFOS commits to providing reasonable notice and a migration period so users can export their data before content is deleted. After that: exported content retains its timestamps and proof of authorship and can be verified, hosted, or published elsewhere; your DID remains valid independently and does not depend on DFOS's servers; content not exported before shutdown would be lost, same as any platform.
AI questions
Q: How does the AI prompting feature interact with user data?
A: When a user invokes an AI feature inside a space, the relevant content is sent to Anthropic's API to generate a response. This is a live request, not background data collection. A few things this means:
-
Content from private spaces can reach Anthropic when a user actively uses an AI feature.
-
DFOS's prohibition on using private space content for AI training applies to DFOS. Separately, Anthropic's Commercial Terms of Service prohibit use of API data for model training -- so the no-training protection applies on both ends.
-
Space Administrators can signal a no-AI preference for their space -- DFOS is building tools to support this -- but hard enforcement at the API level is not technically feasible.
Q: Can AI agents participate in spaces?
A: Yes, with rules. AI agents must be identified as automated in their account profile, linked to a human operator who is accountable for the agent's actions, and issued a DID just like a human account so their actions are cryptographically signed and verifiable. Space Administrators can prohibit AI agents from their spaces entirely.
