Legal
Security & Responsible AI Overview
How we protect systems and data, and the principles behind every AI system we build.
Last updated: 24 September 2026
1. Security baseline
- HTTPS everywhere and encryption at rest where supported.
- Multi-factor authentication for privileged accounts.
- Role-based access control and least-privilege permissions.
- Secrets kept in secure storage — never hard-coded.
- Regular backups with tested restoration.
- Logging and monitoring for unusual activity and failures.
- Dependency updates and vulnerability review.
- Separate development, staging and production environments.
- Immediate access revocation when engagements end.
- A documented incident-response process.
2. Responsible AI principles
- Human review for high-impact communication, financial, legal, health-related, hiring and safety decisions.
- Users can edit, review, pause, correct, export and delete relevant data and automation where feasible.
- Important automated actions are logged for auditability.
- We test for hallucinations, unsafe output, prompt injection, data leakage and bias.
- We describe product limitations honestly.
- Confidence thresholds, fallbacks and escalation paths are designed in from the start.
3. AI and data disclaimer
AI-generated content may be inaccurate, incomplete or unsuitable for a particular situation. Users are responsible for reviewing outputs before relying on them. Our products do not replace qualified legal, medical, financial, security or professional advice.
4. Reporting a vulnerability
If you believe you have found a security vulnerability, email security@yukioshi.com with details and steps to reproduce. Please do not publicly disclose it or access data that isn't yours. We appreciate responsible disclosure and will respond as quickly as we can.