AI and Your Data
Last updated: September 2026
MechCircle is adding features that use AI. This page says exactly what leaves MechCircle when you use one, who receives it, what happens to it afterwards, and what we keep. It is written to be accurate, not reassuring.
AI features are switched off for every company until they are turned on for yours. While they are off, nothing on this page happens. Our general Privacy Policy still applies to everything else.
What gets sent
Only the specific thing the feature needs. If you type a reminder in your own words, we send that sentence. Not your file list, not your contacts, not your drawings, not anything the feature does not need.
When
Only when you use an AI feature. Never in the background, never on a schedule. Each use is one request, made at the moment you make it.
Who processes it
Anthropic, through their API. Anthropic makes the Claude models. We use their commercial API under their Commercial Terms of Service, which is a different arrangement from the consumer Claude app that people use on their phones.
Does it train anything
No. Anthropic's Commercial Terms of Service say:
“Anthropic may not train models on Customer Content from Services.”
Customer Content means what we send them and what comes back. This is the opposite of Anthropic's consumer products, where their Privacy Policy says:
“We may use your Inputs and Outputs to train and improve Anthropic AI models, unless you opt out through your account settings.”
MechCircle is on the commercial API, so that consumer rule does not apply to anything sent through MechCircle. One thing to know: Anthropic can use content that its customer explicitly submits to it as feedback. MechCircle does not send feedback to Anthropic, and there is no button in MechCircle that does.
How long is it kept
Up to 30 days, then deleted. Anthropic's own words:
“we automatically delete inputs and outputs on our backend within 30 days of receipt or generation”
One exception you should know about. If Anthropic's automated safety systems flag something, they may keep it longer. Their words:
“if a chat or session is flagged, Anthropic may retain inputs and outputs for up to 2 years.”
What we record
That a call happened: which company, which person, which feature, how much text went in and came back (counted in tokens, the unit Anthropic bills in), how long it took, and whether it worked.
We do not record what was in it. Not the text you typed, not the answer. The table that holds this record has no column for either, so it cannot be stored by mistake.
What AI can see
Nothing you could not already see yourself. An AI feature runs as you, with your permissions. If you cannot open something in MechCircle, neither can the AI feature.
Two things we are required to tell you
You are using AI. When a MechCircle feature uses AI to produce a suggestion or a result, MechCircle will say so at the point where you see it. Anthropic's Usage Policy requires this of us:
“If model outputs are presented directly to individuals or consumers, you must disclose to them that you are using AI to help produce your advice, decisions, or recommendations.”
Check what it gives you. Anthropic's Commercial Terms require us to tell you that:
“factual assertions in Outputs should not be relied upon without independently checking their accuracy.”
In plain words: treat what an AI feature produces as a draft. A date it reads, a name it picks, a number it fills in can be wrong. Look before you save.
Sources
Every quote on this page is from Anthropic's own published documents, as they read in September 2026:
- Commercial Terms of Service — training, and the accuracy notice
- How long do you store my organization's data? — the 30 days
- API and data retention — the flagged-content exception
- Anthropic Privacy Policy — the consumer rule
- Usage Policy — the AI disclosure requirement
Questions
Ask us anything about this through our Contact Us page. If something here stops being true, this page changes first.