Trust
Technology earns trust through controls, not slogans.
Control areas
Seven public commitments.
Risk governance
Classify use case, users, impact, jurisdictions and prohibited or high-risk concerns before build.
Data protection
Define purpose, data flows, lawful basis and consent responsibilities, retention and deletion requirements.
AI evaluation
Test quality, failure modes, unsafe outputs and regressions against product-specific evaluation sets.
Human oversight
Require review or escalation where errors can materially affect people, money, rights or safety.
Security
Use least privilege, secrets management, logging, backups, dependency review and incident procedures appropriate to scope.
Provider resilience
Avoid unnecessary lock-in; define model and provider fallback when continuity or economics require it.
Transparency
Disclose material AI involvement and limitations where users need it to make informed decisions.
Request our current Trust Overview.
Security reports, architecture details and data processing terms are shared only when appropriate and may require an NDA.