AI Usage

1. What these systems are

IaGenify gives you access to generative models and to an agent that can use tools. These systems produce plausible output; they do not verify it. They can state something false with complete confidence, misread a document, invent a reference, or follow an instruction embedded in material they were asked to read.

Output varies between runs even with the same input. Do not design a process that assumes identical results.

2. Verify before you act

You are responsible for checking output before relying on it, and the level of checking should match what the output is used for.

• Do not treat output as professional advice. It is not legal, medical, financial, tax or safety advice, and it is not a substitute for a qualified person.
• Do not use it unchecked where a mistake affects someone's rights, health, safety, employment, credit or access to a service.
• Check quotations, figures, citations and anything attributed to a named source. A reference that looks real may not be.
• Where the agent used a web page or a document, the link it gives you is the thing to read.

3. What happens to what you send

Your prompt and anything you attach are transmitted to the provider of the model you selected, which processes them to produce the response, under that provider's terms.

We do not use your content to train models.

Do not send material you are not permitted to disclose to a processor — someone else's confidential information, regulated data you have no basis to share, or secrets such as passwords and keys. Where an agent needs a credential, give it the connection rather than the secret: credentials you register are encrypted at rest and used to connect, and are not returned.

4. Agents, permissions and oversight

The agent engine runs in your process. What it may do without asking is decided by the mode you choose and the callback you write, not by the model.

Reading

Never asks. It can still cost credits.

Changing files

Asked, or allowed in automatic mode. Previous versions are kept and deletions go to a recycle bin.

Changing connected data, and DNS

Asked even in automatic mode. Neither can be undone.

Sending, publishing, buying capacity

Asked every time, in every mode, with no way to pre-approve. None of the three can be taken back.

Budget

A credit ceiling stops a run before the next call rather than after the bill.

Set these deliberately. An agent with a broad mode, no budget and wide access will do more than you intended faster than you can watch it.

An agent follows text it reads. Material from a web page, a document or an incoming message can contain instructions aimed at it — treat it as data, and keep approvals for anything that acts outside your own systems.

5. Output, ownership and attribution

As between you and us, output generated for you is yours, subject to the Terms of Service and to any terms of the model provider.

Output is not necessarily original and may resemble material the model was trained on. You are responsible for checking that what you publish or distribute does not infringe someone's rights.

Where you publish AI-generated material, consider saying so. Some jurisdictions require it, and some uses — a likeness, a voice, an endorsement — require consent you must obtain yourself.

6. Uses we do not allow

The Acceptable Use Policy applies in full. In particular, do not use the models or the agent to generate material that sexually exploits children, to impersonate a real person or organization in order to deceive, to produce targeted disinformation, to generate malware or attack tooling, or to make consequential decisions about people without meaningful human review.

7. Automated decisions

If you use IaGenify to make decisions about people, you are the one making them. Where the GDPR applies, a decision based solely on automated processing that produces legal or similarly significant effects is restricted, and you are responsible for the lawful basis, the human review and the information given to the people concerned.

8. Changes

We may update this policy as models and the agent change. The date at the top is the version in force.

The English version of this policy governs; a translation is provided for convenience only.

These documents are written in English, and the English version governs. Questions about any of these reach contact@iagenify.com.