AI Agent Terms

Last updated: September 20, 2026

These AI Agent Terms ("AI Terms") govern your use of TrueHold AI, the AI agent in the TrueHold app, shown as "Agent", "Ask TrueHold AI" and "TrueHold AI Assistant" (the "AI Agent"). They supplement our Terms of Service and Privacy Policy, and words defined there have the same meaning here. Where these AI Terms and the Terms of Service differ on anything specific to the AI Agent, these AI Terms apply. By opening or using the AI Agent you agree to these AI Terms. If you do not agree, do not use it; you can switch it off in the app menu.

1. You are talking to an AI

The AI Agent is software, not a person. Its answers ("Outputs") are generated automatically by artificial intelligence from your questions ("Inputs") and the data described in these AI Terms. No one at TrueHold writes, reviews or approves an Output before you see it, and some answers to common questions are pre-written rather than generated. Outputs are not statements, promises or advice from TrueHold.

2. What the AI Agent does

The AI Agent answers questions about the portfolio you have connected to TrueHold. To do that it can read, for your account only:

• your portfolio summary: net worth, how it splits between wallets, exchange accounts, prediction markets and manual holdings, and how it has changed over time;
• the wallets you added, including their names and public addresses, their holdings and their DeFi positions;
• the exchange accounts you connected: account names, balances, derivatives positions and open orders;
• your transactions and transfers, including fees, your balance history and your performance;
• perpetual futures positions (entry, leverage, unrealized profit and loss, liquidation price) and prediction-market positions.

It can also look up market prices, search the web for public information, and prepare a longer research answer about an asset that combines several AI steps with web searches. You can ask by voice: the recording is converted to text by a speech-to-text provider. Answers can include charts, sources and suggested follow-up questions.

3. What the AI Agent does not do

• It cannot trade, transfer or withdraw, place or cancel orders, or move any asset. It has no access to your funds.
• It cannot connect or disconnect accounts, change your settings or create alerts, and it does not change your portfolio data.
• It never receives your exchange API keys or other credentials. TrueHold never asks for seed phrases or wallet private keys; do not share them in a chat.
• It cannot open websites or links. The only way it reaches the internet is a web search.
• It does not give financial, investment, tax, legal or accounting advice, and it is instructed never to recommend buying, selling or holding an asset, an allocation or a trade.
• It does not make decisions about you that have legal or similarly significant effects.

4. Not advice

Outputs are general information about the data you hold and about markets. They are not financial, investment, tax, legal, accounting or other professional advice; not an offer, solicitation or recommendation to buy, sell or hold any asset or to use any service; and not investment research or an investment recommendation. They do not take into account your objectives, financial situation, knowledge or tolerance for risk. Using the AI Agent does not create an advisory, fiduciary or other professional relationship between you and TrueHold. You alone decide what to do with your assets and you bear all the risk of those decisions. Speak to a qualified professional before making financial, tax or legal decisions. Crypto assets are volatile, and you can lose all of the money you put into them.

5. Outputs can be wrong

AI makes mistakes. Outputs can be inaccurate, incomplete, outdated, inconsistent or biased, and they can sound confident and detailed while being wrong. In particular:

• the AI Agent can misread your question, use the wrong data, miscalculate totals, returns or profit and loss, or describe things that do not exist;
• your data comes from blockchains, exchanges and market-data providers that we do not control, and it can be delayed, missing or wrong;
• web results come from third-party websites that we do not check or endorse, and they can be false, outdated or deliberately misleading;
• the same question can get different answers at different times, and answers can be less accurate in some languages.

Do not rely on an Output as the only basis for any decision. Check anything that matters against the original source (your exchange, a block explorer or your own records) before you act on it.

6. Your responsibilities

You are responsible for your Inputs and for how you use Outputs. You confirm that you have the right to share your Inputs and that they do not break any law or infringe anyone’s rights.

• Do not enter seed phrases, private keys, passwords, API secrets, identity documents, payment card numbers or health information. We are not responsible for any loss that results if you do.
• Do not enter anyone else’s personal data unless you are allowed to share it.
• Verify Outputs before relying on them (section 5), and make your own decisions.

7. What you may not do

You may not use the AI Agent to do, or try to do, any of the following:

• break the law or help anyone else break it, including fraud, market manipulation, money laundering, terrorist financing or sanctions evasion;
• get around its instructions or safeguards, including through prompt injection or jailbreaking, or extract its instructions or configuration;
• access another user’s data, or any data other than your own account;
• send questions or collect Outputs with bots, scripts or other automated tools, or get around usage limits;
• use Inputs or Outputs to build, train or improve an AI model or a competing product;
• sell or commercially exploit Outputs, or present them as advice from TrueHold or as written by a person;
• create content that is unlawful, harmful, harassing or defamatory, or that infringes anyone’s rights;
• probe, test or attack the security of the AI Agent without our written permission.

We may block questions, limit or suspend your access to the AI Agent, or close your account if we reasonably believe you have broken these rules.

8. Third-party AI and data providers

The AI Agent relies on service providers that process data on our behalf:

• The AI model provider: Anthropic’s Claude, run for us in the European Union on Amazon Bedrock. It generates the answers and the follow-up suggestions.
• A speech-to-text provider, OpenAI, which converts voice recordings into text.
• A web search provider, Tavily, which receives the search queries the AI Agent writes. The AI Agent is instructed not to put your holdings into search queries.
• Market-data providers, such as CoinGecko, which supply prices. They receive no personal data.

Some of these providers may process data outside your country, including outside the European Economic Area, with the safeguards described in our Privacy Policy. We choose these providers and may change them at any time. We do not control how their models work, and when a provider is unavailable, the AI Agent or part of it may not work.

9. Your data and your chats

What the AI Agent receives. With every question, the AI Agent receives your question, earlier messages in the same chat and a summary of your portfolio. When it needs more to answer, it retrieves the portfolio details listed in section 2, including wallet addresses. It does not receive your email address, your name or your exchange credentials. Data is sent to the providers in section 8 only to produce the answer or feature you asked for.

What we keep. We store your chats (questions, answers, sources and charts) and usage records (such as the number of questions and their processing cost) so that you can come back to them and so that we can run, secure, support and improve the AI Agent. We do not store your voice recordings; the text of a spoken question is stored as part of the chat. Chats are kept until you delete them or your account. Deleting a chat removes it and its messages permanently. If you rate an answer, we keep the rating together with that question and answer until you delete your account, even if you delete the chat.

Review and monitoring. We may review chats, ratings and usage, by automated means and by our staff, to operate and secure the AI Agent, prevent abuse, answer support requests, fix errors, and evaluate and improve the quality of its answers. We may use questions and answers, with account identifiers removed, to test the AI Agent. Our error-monitoring tools and, if you consent, our product analytics may record technical data about your use of the AI Agent, which can include content shown on your screen.

Training. As our Privacy Policy states, we do not use your personal data to train third-party foundation models.

Our Privacy Policy explains how we process personal data and how to exercise your rights, including access and deletion. For a copy of your chats, write to hello@truehold.xyz.

10. Availability, limits and changes

The AI Agent is a new and evolving feature. We may offer it free of charge or as part of paid plans, and we set limits on questions, voice input, web searches and research answers; the current limits are shown in the app and may change. We may change the models, providers, features, modes and limits, or suspend or withdraw the AI Agent in whole or in part, at any time and without notice. We do not guarantee that the AI Agent will be available, uninterrupted or free of errors.

11. Ownership and licences

TrueHold and its licensors own the AI Agent, including its software, instructions and design. As between you and TrueHold, you keep your rights in your Inputs. You give TrueHold a worldwide, non-exclusive, royalty-free licence, which we may sublicense to our service providers, to host, copy, process and use your Inputs, Outputs and feedback to provide, secure, support and improve the Services, for as long as those purposes require. Subject to these AI Terms, you may use Outputs for your own personal, non-commercial purposes. Outputs may be similar or identical to answers given to other users, and we do not warrant that they are free of third-party rights.

12. Disclaimers

To the maximum extent permitted by law, the AI Agent and all Outputs are provided "as is" and "as available", without warranties of any kind, express or implied, including warranties of accuracy, completeness, timeliness, reliability, merchantability, fitness for a particular purpose, non-infringement, and freedom from bias or offensive content. No Output creates a warranty or commitment that is not expressly stated in these AI Terms or the Terms of Service.

13. Limitation of liability

To the maximum extent permitted by law, TrueHold, its affiliates and their directors, officers, employees and service providers are not liable for any loss or damage arising from or related to the AI Agent or any Output, including trading or investment losses, loss of assets, lost profits, tax consequences, or any decision you make or action you take or do not take based on an Output; and they are not liable for any indirect, incidental, special, consequential or punitive damages. Our total liability related to the AI Agent is subject to the cap in section 10 of the Terms of Service. Nothing in these AI Terms excludes or limits liability that cannot be excluded or limited by law, and if you are a consumer, you keep any mandatory rights that your law gives you.

14. Indemnity

To the extent permitted by law, you will defend, indemnify and hold harmless TrueHold, its affiliates and their directors, officers, employees and service providers against any claims, losses, liabilities, costs and expenses (including reasonable legal fees) arising from your Inputs, your use of Outputs, or your breach of these AI Terms or the law.

15. Changes to these AI Terms

We may update these AI Terms. We will post the new version on this page, change the date at the top, and give additional notice of material changes where the law requires it. If you keep using the AI Agent after an update, you accept the updated AI Terms.

16. General

These AI Terms are governed by the law and dispute provisions in section 13 of the Terms of Service. If any part of these AI Terms is found unenforceable, the rest remains in effect. If we do not enforce a provision, that is not a waiver of it. If we provide a translation, the English version prevails.

17. Contact

Questions about these AI Terms or the AI Agent: hello@truehold.xyz.