Security for a security product.
How Ametriq handles your data: kept separate from every other customer, never used to train AI models, and every decision recorded as evidence.
Your data stays yours
Each customer's policies, evidence records and approval queues are kept separate from every other customer's. We never train AI models on your data.
Where it runs
AIGuard and Spectra run as a service or inside your own cloud account. Hosting region details are shared on request during the pilot. CyberShield runs its checks on each user's own device, so the content it checks is not sent to us to be inspected.
Evidence for every decision
Every allow, block or hold is recorded with the policy that applied, the request it applied to and the risk score, so your auditors can trace each decision.
Compliance status
Our SOC 2 Type II audit is in progress. We will publish the report date here once it is issued. Until then, we share our security package and answer your security questionnaire as part of the pilot.
Our controls are mapped to the OWASP Top 10 for LLM Applications and MITRE ATLAS, the industry's reference lists of AI threats.
Report a security issue
Write to [security@ametriq.ai].
Need a security questionnaire answered?
Request a pilot and we will share our security package with your review board.