Audit Policy

Holding government accountable means being accountable ourselves. This policy explains how our scoring can be reviewed. It applies to every jurisdiction service operated on the platform by Civix IQ, LLC — among them CouncilWatch in Colorado Springs and CommissionWatch in El Paso County.

Its purpose is to provide a transparent mechanism for reviewing the scoring system while safeguarding the integrity, fairness, and intellectual property of the Civix IQ platform.

When audits may be initiated

• Proactively by Civix IQ
• In response to credible, well-documented concerns raised by any party — including elected officials, journalists, institutional partners, or organizations that promote the platform
• As part of a broader third-party review initiative (e.g., academic or civic research partnerships)

Frivolous or duplicative requests may be declined.

What an audit covers

An audit may examine a single government or the platform as a whole. In practice we expect most reviews to look at one jurisdiction at a time, because the questions that prompt them are usually local — a particular vote, a particular classification, a particular member's score.

Because the same rubric and the same pipeline run in every market, a finding in one jurisdiction often applies to others. Where a review identifies a methodology or classification problem rather than a data error, the correction is applied across every government we cover, not only the one that raised it. Those platform-wide corrections are noted in the transparency reporting below.

What we do — and don't — review

We are committed to accuracy and fairness. We review and correct substantive scoring errors, such as when the system misattributes a vote or quote, misclassifies a topic or motion, or lacks key context that would have altered the outcome.

However, we do not review differences of opinion about how the AI interprets governance principles or evaluates the likelihood of a violation. These judgments are made by the system consistently, using a principled, structured rubric developed by Civix IQ. That consistency is essential to the system's purpose and fairness.

Audit structure

Audits are conducted by a neutral third-party analyst or governance expert approved by Civix IQ and bound by NDA, and focus on fair application of the scoring methodology — not political alignment. Scope may include:

• Review of classification consistency (e.g., how "Clear" vs. "Likely" is assigned)
• Targeted review of 3–5 votes or quotes to assess scoring consistency
• Confirmation that the score was calculated based on stated criteria

Confidentiality & IP protection

Auditors must sign an NDA to protect the neutrality of the review process and the integrity of the underlying system. Proprietary formulas, AI prompts, and internal tools are shared only under strict confidentiality. Public statements by auditors must be authorized and limited to procedural outcomes, not technical details.

Costs

Civix IQ may sponsor audits as part of scheduled internal QA or grant-funded research. Challenge-based audits requested by members or third parties may require a formal written challenge, be subject to cost-sharing or fixed audit fees, and be limited in scope to maintain feasibility.

Transparency reporting

We may publish a summary of audit outcomes on the site. Public summaries describe findings (e.g., "Process upheld," "Classification adjustment made") without revealing formulas.

Submit a review request

To request a review, contact us with your name, the evaluation in question, and your reason for the request. Not all requests are eligible for third-party review.