Foundation Model Transparency Index (FMTI)

Type: Transparency benchmark / disclosure scorecard Owner: Stanford HAI Version: Referenced as 2025 update in source Tags: evals transparency benchmarks Created: 2026-04-22 04:22:18Z Last Updated: 2026-04-22 04:22:18Z

Summary

Foundation Model Transparency Index (FMTI) is presented as a framework that scores major AI providers on disclosure across indicators tied to data, safety, governance, and deployment transparency.

Indicators Mentioned in Source

• Training data sourcing and licensing disclosure
• Safety and red-team reporting
• Governance and misuse policy clarity
• Visibility into deployment contexts and downstream use

Findings Referenced

• Average transparency score declined year-over-year.
• Proprietary models were described as less transparent than open-weight counterparts.
• Low transparency reduces accountability and makes independent safety verification harder.

Connections

• Primary source summary: AI Transparency Is on the Decline (Stanford HAI 2025)
• Policy implications: AI Transparency Regulation