AI Terms
What athna AI does with your content, who runs the models, and the promise that nothing trains on your work unless you say so.
Last updated
athna AI is the AI inside athna: one chat over everything in your brand workspace, plus help inside scripting, titles, thumbnails, and the review after a post goes out. These terms say what it does with your content, which companies run the models, and what you're responsible for. They're part of the Terms of Service.
The one promise
Your content doesn't train anyone's models unless you opt in. Not ours, not our providers'. We only use providers whose terms let us keep your content out of training, and we hold them to it. There will be a switch in Settings, off by default, if you want to help improve athna with your content; leaving it off costs you nothing.
What athna AI does
- Reads your workspace to help you. Your posts, scripts, notes, ideas, the stats from your connected accounts, and the reference channels you've added are its context. That's what makes its advice about you instead of a generic creator.
- Coaches from your own results. It looks for patterns in what performed for you (hooks, lengths, topics, when you publish) and uses them when it helps you plan or script.
- Drafts and suggests. Scripts, titles, thumbnail ideas, outlines, cuts from a longer video. You decide what to keep.
- Transcribes. Footage and voice captures are turned into text so you can search them and so the AI can work from them.
- Remembers, in your Vault. Things it learns about you and your brand are stored in your workspace, where you can see and edit them, not inside a model.
Who runs the models
athna AI is built on models from these companies. Each receives only the content a request needs, only when you use a feature, and only to answer that request.
| Company | What it's used for |
|---|---|
| Anthropic | The language models: chat, drafting, coaching, review |
| Voyage AI | Embeddings, the math that lets athna AI search your workspace by meaning |
| Deepgram | Speech-to-text for footage and voice captures |
| fal | Image generation for thumbnails |
They're listed as subprocessors in the Data Processing Addendum. If we add or change a provider, this table changes first, and account holders get an email.
What you're responsible for
- What you publish is yours. athna AI drafts; you publish. Read what it wrote before it goes out. If it says something false, offensive or infringing and you publish it, that's on you the same as if you'd typed it.
- Check the facts. Models are confident and sometimes wrong. Numbers, names, claims about people: verify them.
- Don't use it to break the rules. Everything in the Acceptable use policy applies to what you make with athna AI, including trying to talk it into making something on that list.
- Disclose where you have to. Some services and some countries require you to label AI-made content. Knowing when is your job; we'll make labeling easy where a service supports it.
What it makes is yours
Whatever athna AI generates for you (scripts, titles, thumbnails, cuts) is yours to use, to the extent the law lets anyone own it. We don't claim it, and we don't reuse it for anyone else. Two people asking for similar things may get similar answers; that's how the models work, and it doesn't mean either of you copied the other.
Limits
AI features are included in plans as monthly credits that reset each month; heavy use beyond that is billed as overage packs, as the Terms of Service describe. We may limit unusually heavy or automated use so the service stays fast for everyone.
Where a human is in the loop
No one at athna reads your conversations with athna AI or the content it processes, except: with your permission, when you report a problem and ask us to look; to investigate abuse of the service; or when the law requires it. Error reports sent to our monitoring tools are scrubbed of content first.