Trustworthy & Responsible AI
We use AI to assist reviewers, not to replace them.
Quick Summary
SecurePoint USA uses AI to summarize possible matches for reviewers and, when staff scan an ID, to read the text on it. Name matching itself is deterministic, not AI.People make every decision on a possible match, and decisions and administrative actions are recorded in an append-only audit trail. Our AI practices are built on a foundation of safety, security, and transparency.
Methods
- Name matching runs on deterministic fuzzy matching, not AI
- AI summaries of a possible match for the reviewer
- No AI reads a scanned ID: the visitor types their details
Safeguards
- Human-in-the-loop adjudication
- Append-only audit trail for decisions and outcomes
- AI summaries stored with the provider and model version
- Screening completes without a summary if the AI service is unavailable
Frequently Asked Questions
Is AI required to make access decisions?
No. AI never makes an access decision. A visitor is cleared automatically only when screening finds no possible match, and every possible match goes to an authorized person.
How are AI outputs governed?
An AI summary is added to the screening record for the reviewer to read, with the provider and model version, and it does not clear or hold a visitor. The reviewer's decision and reason are recorded in the append-only audit trail.
Does AI read a scanned ID?
No. A scanned ID photo is kept with the visit and is not sent to AI. The visitor types their details.
Important Disclaimer
AI outputs are assistive signals, not determinations. Compliance and access decisions are solely made by authorized personnel.