Legal

AI Policy

Last updated 2026-10-23.

MarketNewsAPI is operated by DTS Analytics LLC ("DTS Analytics," "Company," "we," "us," "our"). This AI Policy describes how models are used in the Service, what they cannot be relied on for, and what you are responsible for when you use them. It is part of the Terms of Service and should be read with the Privacy Policy.

By calling the API, reading a summary, or connecting an assistant to the MCP server, you acknowledge this policy.

1. What the models do

Two different things are easy to mix up. We do one of them. Your assistant, if you connect one, does the other.

Enrichment of articles

When an article enters the feed, we send the article text to a model that returns:

a short neutral summary, usually one to three sentences;
entities such as tickers, companies, people, and products, which we normalize where we can.

Those fields are attached to the article and returned by the REST API, the WebSocket, and the MCP tools. They are generated once for the article. They are not a conversation with you, and they are not written for your portfolio.

The MCP server

On Pro, including the 7-day trial, you can connect an assistant such as Claude, ChatGPT, Cursor, VS Code, Windsurf, or another client that speaks MCP. The assistant decides when to call our tools. A tool searches the feed or fetches an article and returns the result, including source links, to that assistant. The assistant then writes the answer its own model produces.

We do not operate Claude, ChatGPT, or Cursor. We do not see the rest of your chat except the tool arguments the client sends us, and we do not control the sentence the assistant shows you.

2. Accuracy, omissions, and fabricated text

Models predict likely text. They do not look a fact up in the way a person checks a filing. Even with the article in front of the model:

A summary can omit the point that matters, merge two facts, or state something the article does not say.
An entity or a ticker can be missing or assigned to the wrong company, especially where symbols collide or the article mentions a company in passing.
The same article can be reprocessed if we change models, so a summary can change.
An assistant using the MCP server can ignore a tool result, blend it with other sources, or add claims that are not in the article.

Before you rely on a field, open the original article through the link we return. Before you show a summary to your own users, decide whether your product checks it.

3. No financial, investment, or trading advice

A summary, an entity list, and an answer from an assistant connected to our MCP server are not advice.

They are not investment advice, trading advice, financial advice, tax advice, or legal advice.
They are not a recommendation to buy, sell, hold, or avoid any security.
They do not take account of your finances, your positions, your risk tolerance, or the laws that apply to you. The enrichment model never sees your account.
A positive or negative score is a description of wording in an article. It is not a forecast of a price.
Past articles are not a guide to future prices.

DTS Analytics LLC is not a registered investment adviser, broker-dealer, or fiduciary. No professional relationship is created because a model labeled an article or because an assistant called our server. You are responsible for your own decisions and for any decision your software makes automatically from these fields. If you need advice, ask a licensed adviser.

4. How article text is processed

Enrichment requests go to MiniMax. The default model is MiniMax-M3.1-Flash. We may change provider or model in order to keep the enrichment running. A change of model can change summaries and entity tags.
The payload is the article, not your name, your email, your key, or your billing data.
We use these providers to produce the fields the product shows. We do not authorize them to use your personal information to train their public models. We do not use customer account data or API queries as training data for a foundation model of our own.
Older articles in the archive may not have enrichment yet. A missing summary means the field is empty, not that we judged the article unimportant.

Do not put personal data, account credentials, or customer secrets into a search query or an MCP argument. A query is logged as usage data under the Privacy Policy.

5. Your assistant and your keys

When you connect a third-party assistant, that assistant's provider processes your prompts under its own terms and privacy policy. Review those terms before you connect a client that can see private repositories, files, or other tools beside ours.

You are responsible for the key you give the client, for revoking it, and for the tool calls it makes.
Approve connectors from the dashboard when the product offers that, so the key is one you can identify and revoke.
A client that caches articles is still bound by the Acceptable Use Policy. Connecting a client does not expand the license.
We can suspend MCP access if a client abuses the Service or if your plan no longer includes it. Free does not include the MCP server. The 7-day trial does.

6. Prohibited uses of model output

You agree not to:

Treat a summary, a score, or an assistant's answer as the only basis for a trade, a customer-facing investment recommendation, or a regulatory filing.
Present our output as personalized advice, or as a statement by DTS Analytics about a security.
Remove the source link when you show a summary to an end user.
Use the enrichment pipeline, or a high volume of MCP calls, to reconstruct the model, the prompts, or a copy of the corpus for a competing service.
Submit someone else's private documents for enrichment. Enrichment runs on the news we ingest, not on files you upload. There is no upload feature for that purpose.
Attempt to override safety behavior in a connected assistant in order to produce market-manipulation content, fraud, or other unlawful output that uses our data.

7. No warranty and limitation of liability

Model output is provided "as is" and "as available," without any warranty of accuracy, completeness, or fitness for a purpose. To the fullest extent permitted by law, DTS Analytics is not liable for a decision you or your software make from a summary, an entity, or an answer an assistant produced after calling the MCP server, or for an outage or a change at a model provider.

The liability cap and the exclusions in the Terms of Service apply to every claim about AI features.

8. Reporting a bad field

If a summary is clearly false, a ticker is clearly wrong, or an assistant's use of the feed produced something harmful, email [email protected] with the article identifier or URL, the field you saw, and what the original source says. We use those reports to correct data and to judge model changes. A report is not an admission that we owe you a particular result.

9. Changes

We may update this policy, and we may change models, by posting the update and changing the "Last updated" date. If you keep using the Service after that date, the update applies. A model change that does not change the API contract may happen without a 30-day notice. A breaking change to the API still follows the changelog commitment in the Terms.

10. Contact

Questions about this policy:

DTS Analytics LLC, 1309 Coffeen Avenue STE 1200, Sheridan, Wyoming 82801, USA
Operating as MarketNewsAPI
[email protected]