1. Principles
Anthedon's systems support work that people remain accountable for. AI helps them do that work, but wherever judgment matters, a person makes the call. This policy applies to the AI features in our products and to our own use of AI, on this website and elsewhere in our business.
2. What we commit to
- People decide. When a decision significantly affects someone, such as granting aid, determining eligibility or resolving a complaint, a person makes it. AI can draft, summarize or recommend.
- Clear labeling. We disclose where our products use AI, and drafts produced by AI are labeled so users can check them first.
- No training on your data. Customer content is never used to train AI models, by us, by MAIUMAS or by the model providers we use. Section 3 gives the details.
- Customer settings. Each organization chooses which AI features are enabled, for which users and with what data.
- Audit trail. Every action an AI feature takes in our products is written to the audit log for later review.
- Testing. Before release, AI features are checked for accuracy, bias and potential misuse, and they are monitored once live.
3. Customer data is not used for training
This is a fixed rule of how our products work, not a setting each organization has to find and switch off:
- Never used for training. Customer content is not used to train or fine-tune AI models, whether by Anthedon, by MAIUMAS Ltd, which develops the products we license, or by anyone else.
- Providers bound by contract. We use model providers whose terms prohibit training on the data sent to them through their services.
- Only what the task needs. An AI feature sends a model only the information needed for the request in front of it. Features that draft marketing copy are designed to exclude donor, payment and other personal information.
- Promotion only with approval. A case study, result or quote that names or could identify an organization, including its name or logo, is published only after the organization approves that exact material.
- Changes announced first. If this rule ever changes, any new use would be opt-in only, and we would update this page before the change takes effect.
4. Model providers and data
AI features run through the platform's AI layer, which passes a model only the data needed for the task at hand. Each organization's settings control which model providers may be used, including a private deployment when its own policies call for one. Platform data is hosted by MAIUMAS in Frankfurt, Germany, and a model provider receives only the request it is asked to handle.
5. AI on anthedongroup.com
Apart from the automated security check that protects our forms, this website makes no automated decisions about visitors, and we do not use AI to profile them. AI tools may help us draft pages, and a person reviews every page before it goes live. Anything you send through our forms is handled as our Privacy Policy describes.
6. Rules for people using AI features
You stay responsible for what you do with the output of AI features in our products. In particular:
- check AI output before you rely on it or pass it on to anyone
- follow our Acceptable Use Policy whenever you use these features
- keep health records and other sensitive personal information out of any feature not configured to handle them
- let people know they are dealing with an AI system when applicable law requires it
7. Legal framework
Our AI features are designed to meet applicable U.S. federal and state law, for us and for our customers. That includes the FTC Act's ban on unfair or deceptive practices, which reaches claims about AI, and state AI and privacy laws as they take effect. We do not offer AI for uses U.S. law prohibits.
8. Raising a question or concern
Email [email protected] with any question about how our products use AI, or if you think an AI feature has caused harm or produced a wrong result. We will review it and get back to you.
9. Updates to this policy
As our AI features and the laws that govern them evolve, we will revise this policy, show the new version and date at the top, and log each change below.
Change history
| Version | Date | Change |
|---|---|---|
| 1.0 | November 4, 2026 | First version for Anthedon LLC. |
