El Paso Municipal AI Bias Audit Checklist
El Paso, Texas city officials and oversight teams must ensure municipal AI systems comply with local bylaws, procurement rules, and transparency obligations. This checklist explains records, review steps, stakeholder notices, and escalation paths to detect and mitigate algorithmic bias in city-operated or procured systems. It is designed for compliance officers, legal counsel, procurement, and records staff working within El Paso municipal structures.
Scope & Key Definitions
This audit covers AI systems used to make or materially inform municipal decisions (licensing, permitting, enforcement, benefits, public safety analytics). Define data inputs, model outputs, decision thresholds, and human-in-the-loop roles at the outset, and record versioning and training-data provenance.
Checklist: Governance & Documentation
- Document ownership, procurement contract clauses, and vendor transparency requirements, including rights to audits and source artifacts.
- Maintain a dataset inventory with schema, collection dates, sampling methods, and known biases or gaps.
- Record model versioning, training procedures, hyperparameters, and test results for each deployment.
- Establish routine bias testing (cardinality, disparate impact, false positive/negative rates) and log remediation actions.
- Set periodic review cycles and triggers for ad hoc reviews after incidents or public complaints.
Transparency & Public Records
Publish non-sensitive model descriptions, decision criteria, and summary performance metrics in public-facing records or the city open-data portal where appropriate to meet transparency expectations. For data subject requests and records submissions rely on City of El Paso public information procedures[1].
Penalties & Enforcement
Because El Paso does not publish a dedicated municipal AI ordinance as of the cited sources, specific monetary fines for AI bias or transparency breaches are not specified on the cited pages; enforcement typically follows existing code, procurement remedies, and public-records statutes[2].
- Fine amounts: not specified on the cited page.
- Escalation: not specified; expect contract remedies for vendors, corrective orders for departments, and civil remedies under state open-records law.
- Non-monetary sanctions: orders to disclose, suspension of system use, procurement debarment, or court actions (not specified on the cited page).
- Enforcer: responsible departments include Procurement, City Attorney, and City Clerk for records and complaints; submit complaints or information requests via official City channels[1].
- Appeals and review: governed by existing administrative appeals and judicial review; specific time limits are not specified on the cited page.
Applications & Forms
No El Paso-specific "AI system registration" form is published on the cited pages; use standard public-information request forms and procurement contract remedies where needed[1].
Technical Review Steps
- Reproduce results on hold-out sets and check subgroup performance by protected class proxies.
- Verify input data lineage and retention policies comply with city records rules.
- Test for dataset shift, label bias, and demographic imbalance; require retraining or threshold adjustments if disparities exceed policy thresholds.
Action Steps for Officials
- Include AI audit rights in new vendor contracts and add transparency clauses to renewals.
- Adopt scheduled reviews and incident-triggered audits with clear ownership.
- Provide a public contact and complaint intake pathway for residents to report potential bias; use official City Clerk request pages for records and complaints[1].
FAQ
- Who enforces AI transparency and bias rules in El Paso?
- Enforcement is handled through existing departmental oversight (Procurement, City Attorney, City Clerk) and contract remedies; specific AI enforcement provisions are not published on the cited pages.
- Can the public request model details?
- Yes — submit a public information request via the City Clerk procedures to request non-exempt records and summaries of model performance[1].
How-To
- Identify all municipal AI systems and responsible program managers.
- Collect documentation: data inventories, model artifacts, contracts, and vendor attestations.
- Run bias and performance tests; document disparities and planned mitigations.
- Report findings to legal and procurement, implement remediations, and publish non-sensitive summaries to the open-data portal when possible.
Key Takeaways
- Create contractual audit rights and transparency clauses for vendors.
- Maintain dataset and model inventories to support timely reviews.
- Use City Clerk public information channels for records and complaint intake.
Help and Support / Resources
- City of El Paso Code of Ordinances
- City Clerk - Public Information Request
- El Paso Open Data Portal
- City Purchasing / Procurement