# AI safeguards and daily limits

> What OpenATS’s AI never considers, how it treats sensitive and untrusted text, and how the daily AI allowance works.

OpenATS’s AI is advisory only: it never decides, rejects or advances anyone. It never searches, scores or tags on who someone is, hands sensitive messages to a person, and ignores instructions hidden in resumes and documents. Each workspace has a daily AI allowance, shown in **Settings → Usage**, that resets at midnight UTC.

*Search & AI · Last updated 2026-11-11*

---

## Never about who someone is

Search terms, Ask questions, match criteria, tag rules and AI tags all leave out anything about who someone is, and the answers are checked again on the way back. That covers age, gender, race or ethnicity, religion, disability or health, family status, national origin, citizenship or immigration, sexual orientation, veteran status, political views, criminal history and salary history, and things that stand in for them, such as graduation years, “native speaker”, “culture fit” or gaps in a career.

## Sensitive messages go to a person

In OpenATS Cloud, an email or document that asks for an accommodation, shows distress, mentions legal action, a visa or a negotiation goes to a person on your team. Other emails to that candidate wait until it’s handled. For an accommodation, a visa or a negotiation, the card offers to draft a reply, written only when you ask and sent only after you’ve edited it. It never feeds evaluations or what OpenATS learns.

## Text addressed to an AI

Resumes and documents sometimes hide instructions for AI tools. OpenATS takes them out before anything reads the file, and tells the people handling the candidate: **The resume has text addressed to an AI… OpenATS doesn’t follow it.**

## The daily allowance

AI use is shared out per workspace per day (UTC). **Settings → Usage** shows **Your AI today** and the workspace’s use, with **By person** for admins. This daily share is separate from the AI actions your plan includes each month; see [AI actions, the limit and usage credits](https://openats.ai/docs/billing/ai-actions-and-usage-credits).

- Background reading stops first, partway through the day’s allowance, so drafts, summaries and answers always have the rest.
- Each person has a share of the day. When yours is used up, OpenATS says **You’ve used today’s share of OpenATS’s AI; it resets at midnight UTC.**
- At 100%, AI pauses for everyone in the workspace until midnight UTC. Everything that doesn’t need AI keeps working.

## Questions about AI safeguards and daily limits

### What happens when AI is off?

The AI features don’t appear, and everything else works as usual: search, the pipeline, email and imports don’t need AI.

---

Still need help? Write to the OpenATS team at https://openats.ai/support (or support@b150.ai). We reply by email.

Source: https://openats.ai/docs/search-and-ai/ai-safeguards-and-limits
OpenATS help center: https://openats.ai/docs