---
title: "The U.S. Government AI Wastes Billions on Wrong Priorities"
description: "5 source-backed AI claims from The U.S. Government AI Wastes Billions on Wrong Priorities, with key statistics, context, and the original AI Adopters Club…"
url: "https://kbanc.com/claims-library/us-government-ai-wastes-billions-on-wrong-priorities"
source: "https://aiadopters.club/p/the-us-government-ai-wastes-billions"
date: "2025-08-07"
topics: ["measurement", "strategy", "implementation"]
generated: "2026-08-31"
---

# The U.S. Government AI Wastes Billions on Wrong Priorities

By Kamil Banc | August 7, 2025

## Claims

1. **IRS AI Fraud Recovery** (source summary): The IRS deployed AI fraud detection that recovered $375 million but delayed $20 billion in legitimate refunds.
2. **81% False Positive Rate** (source summary): A Treasury Inspector General report found the IRS fraud detection filters had an 81% false positive rate.
3. **USCIS Processing Time Cuts** (source summary): USCIS cut average case processing times from 10.5 months to 6.1 months using AI systems.
4. **DOJ System Fragmentation** (source summary): The Department of Justice maintains twelve license plate reader systems and nine separate AI transcription systems.
5. **Measure Broader AI Costs** (Kamil's interpretation): Organizations should measure stakeholder burden and equity of outcomes, not just time saved and costs reduced.

## Evidence

### Quote
> "The AI didn't eliminate the work of fraud detection, it massively amplified it while making innocent taxpayers pay the price." - Kamil Banc

### Key Statistics
- **81%**: False positive rate of IRS fraud detection filters, per a Treasury Inspector General report
- **$20 billion**: Legitimate refunds delayed to compliant taxpayers while the system protected $7.6 billion in revenue
- **10.5 to 6.1 months**: USCIS average case processing time reduction achieved across a record 10.9 million cases
- **$2.60 per $1**: Legitimate transactions wrongly flagged for every dollar of actual fraud caught by IRS AI

## Context
The article synthesizes government audit findings, agency performance data, and think-tank analyses to expose the 'AI efficiency trap,' where narrow efficiency metrics mask costs shifted onto stakeholders. Author Kamil Banc combines his personal experience navigating USCIS systems with documented cases from the IRS, USCIS, and Department of Justice. Practitioners should audit whether their AI success metrics capture downstream burdens such as false positive remediation, applicant preparation costs, or inequitable outcomes. The recommended approach pairs efficiency measures with burden and equity metrics, supported by human-in-the-loop review to catch algorithmic failures before they compound.

## Source
- Original: [The U.S. Government AI Wastes Billions on Wrong Priorities](https://aiadopters.club/p/the-us-government-ai-wastes-billions)
- Cite: kbanc.com/claims-library/us-government-ai-wastes-billions-on-wrong-priorities

## Primary Evidence
- [The Treasury reported recovering $375 million](https://home.treasury.gov/news/press-releases/jy2134) (home.treasury.gov; supports claim 1)
- [A Treasury Inspector General report found](https://www.taxpayeradvocate.irs.gov/wp-content/uploads/2020/07/ARC18_Volume1_MSP_05_FalsePositiveRates.pdf) (taxpayeradvocate.irs.gov; supports claims 1, 2)
- [cutting average case processing times from 10.5 months to 6.1 months](https://www.forbes.com/councils/forbesbusinesscouncil/2025/05/20/the-role-of-ai-in-uscis-key-facts-and-implications/) (forbes.com; supports claim 3)
- [A Council on Foreign Relations analysis](https://www.cfr.org/blog/ai-federal-government-fragmented-reality) (cfr.org; supports claim 4)