Terms of service.
The terms between BotShield and the business that uses the BotShield Console: the BotShield Gate, linked accounts, Agents Ask, the SDK and the API. Three documents on one page, each short enough to read — the terms, the acceptable use policy, and the service level agreement.
Version 1.0 · Effective October 12, 2026 (replaces May 15, 2026)
Who these terms are between
These terms are between BotShield, Inc. (“BotShield”, “we”) and the business that creates an organization in the BotShield Console (“you”). They cover the BotShield Gate, linked accounts, Agents Ask, the Console, the SDK, the API and the documentation. If you sign an order form with us, the order form wins where the two differ. People who verify with BotShield are covered by the BotShield app’s own terms, summarized at the end of this page.
1. What you are buying
The BotShield Gate. You place a gate on your own site, app or agent. When a person taps it, they prove with their own passkey on their own phone that a human is there, and you receive a signed verdict: pass, or Unavailable, with a reason when it is Unavailable. A Human Gate asks whether a human is here. An Age Gate asks whether this person is over 13, 18 or 21, and answers only that. You choose each gate’s policy — how many times each person may pass in a window, and how many days their BotShield ID must have been real — and we apply it.
Linked accounts. You may offer Secure your account with BotShield on your own surface. A person links one of your accounts to their BotShield ID with their passkey. You receive one opaque handle for that account and the events that follow; the handle is different on every business and cannot be joined across businesses.
Agents Ask. You register agents. An agent proposes an act; the person confirms or denies it on their own device; you receive a signed proof of resolution for either answer, and you carry out the act, or don’t.
Sandbox and production. The sandbox is free, on from your first sign-in, with test keys that work from any origin. Production requires a verified domain, a verified business and a Developer passkey (section 3). The documentation defines the service in detail and is part of these terms.
2. Plans, the meter and your bill
The plans and prices on our pricing page are part of these terms.
How the meter works. Verifications and Agents Ask requests are counted per month and reset on the first of the month. Past your included amount they are billed at your plan’s rate on your next invoice. A person at your gate is never refused because of your plan. Gates and linked accounts are caps: at the cap, the next placement or link is refused until you upgrade. Free never bills: past its included verifications, people keep passing at your gate and you are asked to pick a plan.
Paying. Monthly plans bill in advance each month; annual plans bill in advance for the year at the annual rate. Overage bills in arrears. You pay by card through Stripe, or by invoice on Enterprise. Fees are non-refundable except where these terms say otherwise. You are responsible for taxes other than our income tax. Invoices unpaid 30 days after the due date earn 1.5% a month.
Changing plans. Upgrades take effect at once, prorated. Downgrades and a switch to Free take effect at your next renewal date; on Free, one production gate stays on and you choose which. Cancelling keeps your plan until the end of the period you paid for.
Changing prices. We give you 30 days’ written notice before a price increase takes effect, and an annual plan keeps its price until it renews.
3. Production is earned
Four things unlock production: a domain you verify with one TXT record; a business review — legal name, EIN, website and intended use, reviewed by a person at BotShield, usually within a day; a Developer passkey, with which production keys and gates are confirmed; and a plan (Free counts). The review’s outcomes are in review, verified, needs one more thing, and rejected; we tell you which, and what to add. We may decline production, or withdraw it with notice, for a use that breaks the acceptable use policy; the sandbox stays open to you regardless.
4. What a verdict is, and whose decision follows
A pass tells you a human completed the ceremony. It does not tell you who they are, and what you do with a pass or an Unavailable is your decision. We verify people; you enforce on your platform. A verdict is signed with our published key and works once; you verify it yourself.
An Age Gate returns an age predicate — over 13, 18 or 21 — and its assurance level (declared, guardian, verified), composed from the age signal the person’s platform already holds. A self-declared age never passes; a parent’s declaration passes only the 13 gate; 18 and 21 pass only on the platform's confirmed grade. You may set a higher bar than ours; you may not set a lower one. The Age Gate is a method that supports your age-assurance obligations and gives you a signed, dated record of every check. It is not a certification, not a guarantee, and not a substitute for your own assessment of what your law requires. Where your law requires more than the platform’s signal provides, you must add it. Our age assurance statement is part of these terms.
A proof of resolution records that a person deliberately confirmed or refused; you carry out the act, and we do not. We state the strength of the check, never an absolute: a live-human check has a residual error rate, and one person, one account is uneconomical to defeat, not impossible.
5. Your data, your users’ data
Our privacy policy says what we hold and for how long; our data processing addendum governs the references you send us about your users, for which you are the controller and we are your processor. In short: your users’ references are stored as keyed hashes we cannot reverse; your logs are kept 90 days; we never sell, share, profile or train on your users’ data, and we cannot tell you who they are. You must have the right to send us the references you send, and you must tell us when a user’s primary credential changes on your platform so we can end that link.
6. Your account and your keys
Your organization’s admins are responsible for who is on your team and what they do. Keys and the Developer passkey are yours to keep secret; rotate a key you believe is exposed, and tell us. We may act on instructions from anyone who holds your keys.
7. Acceptable use
The acceptable use policy below is part of these terms. Breaking it is a material breach.
8. What we promise, and what we don’t
We promise the service will perform materially as the documentation describes, and that we will not materially reduce its functionality or security during a paid term. If it doesn’t, tell us; we fix it, or you may terminate and we refund the unused part of what you paid. Beyond that, the service is provided as is, without other warranties, express or implied, including merchantability, fitness for a particular purpose and non-infringement. Beta features are marked, may change or end, and carry no promise.
9. Liability
Neither of us is liable to the other for indirect or consequential loss, including lost profits or lost sales on an Unavailable. Each side’s total liability under these terms is capped at the fees you paid us in the 12 months before the claim, or $100 if you paid nothing. The cap does not apply to a party’s fraud, to a breach of section 5 or the acceptable use policy, to payment obligations, or to the indemnities in section 10.
10. Indemnities, both ways
We defend you against a claim that the service, used as documented, infringes someone’s intellectual property, and we pay what a court awards or we agree. You defend us against a claim arising from your data, your use of the service, your platform, or your breach of law or of the acceptable use policy. Each side tells the other promptly and lets it run the defence.
11. Service levels
The service level agreement below applies to the Scale and Enterprise plans.
12. Suspension, ending, and getting your data back
Suspension. We may suspend at once for a security risk or an acceptable use breach, and we tell you why. For non-payment, we tell you in writing and suspend 10 days later if it is still unpaid. For any other breach, you have 30 days to cure after we tell you.
Ending. You may delete your organization at any time, confirmed with your Developer passkey. Either side may end these terms for an uncured material breach. If you end them because of our breach, we refund the unused part of what you paid.
Your data afterwards. For 90 days after your organization ends you can export your logs and your registry from the Console. After 90 days we delete your organization’s data, except invoices, which we keep as tax law requires.
13. Changes to the service and to these terms
We may improve the service at any time. We give 60 days’ notice before a change to the API that is not backwards compatible, and we keep the old names working for that window. We give 30 days’ written notice before a material change to these terms; the change takes effect at your next renewal unless the law requires sooner.
14. Support
Community support on Free; email support on Starter; priority support on Growth, Scale and Enterprise. The Console’s Support page opens a thread with your organization attached, and replies land in your inbox there.
15. The rest
These terms, the acceptable use policy, the service level agreement, the privacy policy, the data processing addendum, the pricing page and the documentation are the whole agreement. They are governed by the laws of the State of Delaware, and any dispute goes to the state or federal courts located in Delaware; neither side asks the other to arbitrate. Neither side may assign except to a successor of its business. Notices to you go to your organization’s admins; notices to us go to BotShield, Inc., Denver, Colorado, support@botshield.ai. If a clause is unenforceable the rest stands.
Acceptable use policy
Part of the terms. One behaviour per line. Breaking any of them is a material breach. You may not:
If you see a use that breaks this policy, tell us at support@botshield.ai. We investigate, we may suspend the gate or the organization involved, and we tell the business what we found.
Service level agreement — Scale and Enterprise
What is measured. The availability of the verification API — the verify and evaluate endpoints and the key-set endpoint — each calendar month, measured from our monitoring across regions. An incident counts when the API returns errors or no response for at least five consecutive minutes. Scheduled maintenance announced 72 hours ahead, your own network, and a breach of the acceptable use policy are excluded.
The commitment. 99.9% monthly availability on Scale. Enterprise per contract.
The credit. If availability in a month falls below 99.9%, you receive a credit of 10% of that month’s plan fee. Credits apply to a future invoice, cannot exceed one month’s fee, and are the only remedy for an availability shortfall.
How to claim. Open a support thread within 30 days of the end of the month, with the times you saw. We answer with our measurement and apply the credit on the next invoice. Current and historical availability is on status.botshield.ai.
The person’s terms, in short
People who verify with BotShield accept the BotShield app’s terms: sign in with Apple or Google only; the passkey is theirs; BotShield verifies human, never who; they may unlink any account and delete their account at any time; 13 and older.
Questions about these terms: support@botshield.ai. Our privacy policy and our age assurance statement stand beside this page. These are plain-language terms, and they are the ones we mean.
BotShield runs in your browser.
Sign in with Apple or Google, add a passkey, and approve what your agents ask — the same account the phone app will use.
Open app.botshield.ai →