Does it apply to you?
If you use AI to assess people’s creditworthiness, or to price life and health insurance, the Act treats that use as high-risk. Other AI agents may carry lighter duties, but the same evidence still helps with supervisors and audit.
Key duties and the evidence Verus produces
| Duty | What it asks | Evidence from Verus |
|---|---|---|
| Record-keeping (Article 12) | Log events automatically so the system can be traced | A signed record of every decision |
| Human oversight (Article 14) | Let people oversee, override and stop the system | Approval records and a stop switch with its own log |
| Risk management (Article 9) | Find and manage risks across the system’s life | Reliability scores and alerts when behaviour changes |
| Transparency (Article 13) | Help deployers understand and use outputs | Plain-language decision reports |
| Accuracy and robustness (Article 15) | Perform consistently and resist misuse | Per-decision scores and blocked actions |
If you deploy AI built by others
Banks and insurers often deploy AI from vendors. Deployers of high-risk systems must use them as instructed, assign human oversight, monitor them and keep the logs the system generates. Verus gives you that record whoever built the model.
Where to start
- List your AI uses and mark which could be high-risk.
- Pick one workflow, such as credit or AML, to evidence first.
- Turn on logging and oversight before you scale.
- Review the evidence with compliance and legal.
General information, not legal advice. Application dates are phased; confirm the current timeline with your counsel.