Guiding Responsible AI: Sixfold’s 2026 Framework

Why did we develop this report?

2026 was supposed to bring clarity on how AI should be governed. Instead, deadlines moved, laws changed, and governments went in different directions. The questions underneath all of it did not change. Is the system reliable? Can people understand and challenge its work? Who is accountable?

This report answers those questions in practice. It covers how Sixfold governs its underwriting AI today, and what expands as that AI becomes agentic and takes on more of the work on a submission. Judgment, authority, and accountability stay with people.

Who is it for?

Our customers, and any insurance carrier or underwriting team evaluating or using AI in underwriting. The framework applies to both P&C and L&H.

What's in the report?

• Human Oversight
• Bias Prevention & Fairness
• Model Testing & Validation
• Security & Robustness
• Data Governance