HONEST PRIVACY

Not "privacy-focused."
Honestly private.

We're not going to claim things we can't prove. Here's exactly what we do, what we don't do, and how you can verify what's verifiable.

WHY WE BUILT THIS

People are scared. And they should be.

Every AI company says "we care about your privacy." Most of them are simultaneously training on your data, selling insights to advertisers, and writing privacy policies specifically designed to make that legal.

AI51 exists because the people who need it most — people in crisis, people dealing with debt, mental health, trauma, legal threats — those people cannot afford to have their most vulnerable moments become someone else's dataset.

So we built the architecture around that. Not as a feature. As the foundation.

THE TRUST LADDER
"We don't sell your data"Promise
"Our privacy policy says..."Legal text
"We're SOC2 compliant"Audit
Honest architecture + verifiable third-party policiesWhat we actually do
THE ARCHITECTURE — TAP EACH LAYER
HTTPS TRANSPORT
Industry-standard TLS. Your data is encrypted in transit.
BRAVE SEARCH API
Independent index. Zero data retention. No Google. No Bing.
WE DON'T STORE CONVERSATIONS
Your messages are not saved to a database after your session.
IPFS KNOWLEDGE LIBRARY
Content-addressed. Tamper-evident. Survives AI51 disappearing.
WHAT WE DON'T DO

Simple list. No asterisks.

We don't train on your conversations
Your messages are not fed back into any model. We use third-party AI inference (Base44) — we don't control the weights, and we don't contribute your data to them.
We don't sell your data
No advertising. No data brokers. No "anonymised insights" packages. AI51 makes money from subscriptions, not surveillance.
We don't store your conversations long-term
Sessions are not archived to a searchable database. We have no conversation history we could hand to anyone.
We don't use dark patterns
No pre-ticked consent boxes. No privacy policy designed to be unreadable. No "personalisation" that means profiling.
WHAT WE CAN'T PROMISE

Honesty includes the limits.

Can you mathematically prove you don't log?
No. That would require you to audit our infrastructure directly. We're telling you what we do as a design commitment, not a cryptographic proof. We think honesty about that gap is more trustworthy than a bold claim we can't back.
What about Base44 — the platform we run on?
Base44 processes requests as part of serving the app. We've chosen them because their terms support our privacy commitments. But we're honest: there's a layer we don't fully control.
What if governments compel you?
We would comply with lawful orders — but we have very little data to hand over. No conversation archive, no behavioural profiles. That's a real protection, not a legal argument.
THE BIGGER REASON

Privacy is how we show we care.

AI51 is built for people navigating hard things. Legal threats. Mental health crises. Debt. Family breakdowns. The technology exists to help — but only if people trust it enough to be honest with it.

If someone doesn't feel safe being real with AI51, we've failed. Privacy isn't a compliance box. It's the thing that makes the whole project possible.

Built for people who needed help and had nowhere safe to ask.

Honest about what we do. Honest about what we can't prove.

Verify what's verifiable. Trust us on the rest — or don't use it. That's a fair trade.