Public Policy

Responsible AI Policy

What is the Responsible AI Policy?
This policy defines the 5 core criteria—Rate Limiting, Model Audits, Human-in-the-Loop, Red Teaming, and Data Transparency—used to calculate the Responsibility Score and certify domains.

Our Principles

Transparency

All claims are verified with sources. Every fact includes methodology and verification dates.

Fairness

Audit criteria are objective and publicly documented. All organizations are evaluated equally.

Sustainability

Promoting practices that reduce environmental impact and ensure long-term viability.

Audit Criteria

Organizations are evaluated on five key areas. A score of 60 or higher qualifies for the badge.

Rate Limiting

Implementing rate limits to control compute usage and prevent runaway costs.

20 points

Tips to meet this criterion:

Set per-user and per-endpoint limits
Implement exponential backoff
Monitor and alert on limit breaches

Model Audits

Regular audits of AI outputs for bias, accuracy, and quality issues.

25 points

Tips to meet this criterion:

Establish baseline performance metrics
Test with diverse input datasets
Document and address findings

Human-in-the-Loop

Human review for critical or high-stakes AI decisions.

25 points

Tips to meet this criterion:

Identify high-stakes decision points
Create review workflows
Train reviewers on edge cases

Red Teaming

Testing AI systems for vulnerabilities and failure modes.

15 points

Tips to meet this criterion:

Conduct adversarial testing
Test prompt injection resistance
Document and patch vulnerabilities

Data Transparency

Transparent data practices with minimal PII and no unnecessary sharing.

15 points

Tips to meet this criterion:

Minimize PII in prompts
Anonymize where possible
Clear data retention policies

Badge Eligibility

Complete the audit form

Answer questions about your AI usage, data practices, and responsibility measures.

Receive your score

Get an automatic score based on the criteria above. Minimum 60 points for eligibility.

Display your badge

Eligible organizations receive embed code for their website plus a React component.

Ready to get certified?

Start your free audit and find out if you qualify for the Responsible AI badge.

Our Principles

Transparency

All claims are verified with sources. Every fact includes methodology and verification dates.

Fairness

Audit criteria are objective and publicly documented. All organizations are evaluated equally.

Sustainability

Promoting practices that reduce environmental impact and ensure long-term viability.