Terms of Use & Data Consent

Hero Run lets you run AI models and pay with the $HERO token on the Base network. Trading fees from $HERO fund the training of open-source AI models. Using the app means you accept these terms.

When you run a model, your wallet transfers $HERO to the collection wallet on Base. On-chain transactions are irreversible. We cannot refund, reverse, or recover a payment. You are responsible for your wallet, keys, and gas.

$HERO is a utility and community token. It is volatile and can lose all value. Nothing here is financial, investment, legal, or tax advice. Prices, projections, and scenarios shown in the app illustrate the model; they are not predictions or promises of value. Only spend what you are prepared to lose.

Models are provided by third parties through OpenRouter. Outputs are generated automatically and may be inaccurate, offensive, or unsuitable. We make no warranty on any output. You are responsible for how you use what you generate, and for complying with the underlying model providers' terms. Do not submit unlawful content or others' private or infringing material.

This is the part that needs your active consent. If you opt in, the prompts you submit and the outputs the models return may be stored and used to build an open dataset that trains open-source AI models. This is off by default and only happens when you have explicitly agreed.

Your public wallet address and on-chain transactions are, by nature, public. Do not put private, personal, or confidential information in a prompt. You can ask us to remove your contributed examples where technically feasible.

The service is provided "as is," without warranties of any kind. To the maximum extent permitted by law, we are not liable for lost tokens, failed transactions, downtime, model errors, or any indirect or consequential damages.

We may update these terms as the product evolves. Continued use after a change means you accept it.

This section applies only if you contribute to The Big Distill. The Big Distill lets you distill research artifacts from your own agent memory, decrypt them locally in your browser, redact them, and submit them into an open research corpus that feeds Hero Run's training work. Decryption and the mandatory redaction step happen on your device. Only the specific items you select and choose to submit leave your browser.

Ownership affirmation. Before you submit, you sign a message with your wallet affirming that you own or have the right to license the content, and that you have removed third-party personal data and secrets such as keys, account numbers, and passwords. That signature is stored as a consent record bound to the exact content you submitted.

License grant. By submitting, you grant Hero Run a license to store, process, redact, and use the accepted content to build and improve the open research corpus and the models it trains, under the license identified as hero-open-corpus-1.0. You represent that this grant does not infringe anyone else's rights, including the terms of any model provider whose output you include.

Removal and takedown. You can ask us to remove a contributed item, and we will remove the stored item where technically feasible. Be aware of the honest limit: once an accepted item has already influenced a training run or been merged into research notes, removing the stored item cannot undo that prior influence. We keep the corpus off-chain and mutable precisely so removal and redaction remain possible.

Payment. The current version pays nothing for contributions. Any future payout is not live, is not promised, and would only follow legal review. Do not contribute expecting compensation. Contribution is framed as delivering accepted research work into an open corpus, not as an investment in any token.

Acceptance is provisional. An accepted item has passed automated format, duplicate, secret, and personal-data checks plus a model review. That is a quality signal, not proof that it measurably helped. Do not submit unlawful content, other people's private or infringing material, secrets, or content designed to poison a research process.