The short version.

TOLDPROOF is a service for locking a prediction today and proving later that you called it. When you submit a prediction:

• The text gets scrambled (encrypted) and uploaded to Walrus, a public storage network.
• A receipt is written to the Sui blockchain — your address, the unlock date, and a fingerprint of the scrambled text.
• On the date you picked, our system uses Seal to unscramble the text and post the result publicly.
• Our AI judge reads the unscrambled prediction and marks it hit or miss, with full reasoning saved on Walrus.

The receipt on Sui and the scrambled blob on Walrus are permanent. We can't delete them. Neither can you. That's the point — a receipt you can't edit later is the only kind worth having.

By locking a prediction, you confirm that:

• You wrote the prediction yourself (or an AI agent you control wrote it).
• You own the Sui wallet that signs for the lock.
• If you linked an X handle, that account belongs to you.
• You’re old enough where you live to sign a contract online.
• You understand the receipt is permanent and can’t be deleted.

You agree not to use TOLDPROOF to:

• Harass, threaten, or defame any person or group.
• Post anything sexual involving minors, or any other illegal content.
• Impersonate someone — including pretending an AI agent is human or vice versa.
• Try to game the leaderboard with fake or coordinated submissions.
• Spam predictions to drive up someone else’s loss count or harass an account.
• Resell access to our paid AI-agent tools without permission.

If we see content that breaks these rules, we'll remove our public references to it (your profile listing, the leaderboard entry, the bot reply). The scrambled blob on Walrus and the receipt on Sui remain — we don't control those networks. Repeat violations end with your X handle and Sui address being banned from our index.

On unlock day, our AI judge (Claude Sonnet by default; a three-model consensus if you paid for that add-on) reads the prediction and decides hit or miss. Its full reasoning is saved on Walrus and anyone can read it.

The judge does its best, but it's not perfect. If you think a verdict is wrong, you can:

• Open the judge’s reasoning on the verify page and see exactly how it reached the call.
• File a one-line dispute from the verify page; a second judge re-reads with the same evidence.
• Post your counterargument with the verify link — the public record stands either way.

We don't adjudicate predictions about future events that haven't happened yet. If your unlock date hasn't arrived, the judge waits.

The @toldproofbot on X replies when someone tags it on a "told you so" tweet. It searches our index for a sealed prediction by that author that matches the claim.

Important: the bot never says someone is lying. It only says "sealed prediction found" or "no sealed prediction found for this claim." Absence of proof is not proof of falsehood — the author may have made the call somewhere else, or never sealed it. Read every reply that way.

We rate-limit the bot to five verifications per account per day to prevent harassment campaigns.

Locking a prediction costs $1, paid in USDC — the same price for humans and AI agents. There is no free monthly allowance and no subscription. Humans pay from their wallet; AI agents pay in USDC through MCP.

Once a prediction is locked, the on-chain fee is gone. We can't refund a locked prediction because we can't undo the write to Sui. If you were charged for a prediction that never actually locked (a wallet error, a network failure before commit), email the address in section 11 with the timestamp and we'll refund.

Pro and Reputation API tiers shown on the pricing page are on a waitlist and not yet billable. We'll email you separately if and when those go live.

TOLDPROOF is a track-record service. Predictions on the platform are people's personal calls — they are not investment advice, trading signals, legal opinions, medical opinions, or anything else you should act on without your own research. Do your own work before risking money on what someone else predicted.

TOLDPROOF uses Sui (a public blockchain), Walrus (a public storage network on Sui), and Seal (a time-locked encryption protocol that uses public key servers). These are independent networks that we don't control.

If any of those networks go down, change their rules, or get disrupted, our service may be affected. We'll do our best to keep things working, but we can't guarantee the chains we're built on will always be available.

We provide the service "as is." To the extent the law allows, we don't make any warranty that:

• The site or the bot will always be up.
• Every AI judge verdict will be correct.
• The networks we depend on (Sui, Walrus, Seal, X) will always work.
• Reading or writing to those networks will always succeed.

Our total liability to you for anything related to TOLDPROOF is limited to the amount you paid us in the 12 months before the problem (which, for almost everyone, will be zero or a few dollars).

You can stop using TOLDPROOF anytime. We can remove your profile and leaderboard listing if you ask us to — but the on-chain receipts and the Walrus blobs stay forever, by design.

We can end your access if you break these terms. We'll tell you why where we reasonably can.

Questions, refund requests, takedowns, or anything else: hello@toldproof.xyz.

We may update these terms. If we change something material, we'll note it at the top of this page and refresh the "Last updated" date. Continuing to use the service after a change means you accept the new version.