Federal AI Intelligence · U.S. Government AI Observatory
Custom inventory report
Filter: failsafe = blank
Key findings
- 3,320 use-case records from 40 agencies: 848 deployed (26%), 408 pilots, 1,415 in development or acquisition, 311 retired, 338 with no stage reported.
- 317 records are designated high-impact; 0 of them (0%) answered at least one minimum-practice question.
- HHS, NASA, VA report 1,239 of 3,320 use cases (37%).
- 26% of reported deployed use cases (219 of 848) are classified as generative or agentic AI.
- Of 101 deployed high-impact use cases, 0 report a completed AI impact assessment and 101 leave every minimum-practice field blank.
- Interpretation: Blank practice fields are missing evidence, not evidence that a safeguard is absent. They mark where follow-up questions to the agency are most useful.
- Sourcing is reported for 40% of use cases. Where reported, 68% involve in-house development.
- 338 use cases (10%) have no reported development stage.
Lifecycle
| Stage | Records | Share |
|---|---|---|
| Deployed | 848 | 26% |
| Pilot | 408 | 12% |
| Pre-deployment | 1,415 | 43% |
| Retired | 311 | 9% |
| Not reported | 338 | 10% |
AI technology
| Classification | Records | Share | Deployed |
|---|---|---|---|
| Generative AI | 755 | 23% | 200 |
| Classical / predictive ML | 1,010 | 30% | 296 |
| Natural language processing | 421 | 13% | 173 |
| Computer vision | 247 | 7% | 121 |
| Agentic AI | 114 | 3% | 19 |
| Reinforcement learning | 12 | 0% | 5 |
| Other | 99 | 3% | 31 |
| Not reported | 662 | 20% | 3 |
Agencies
| Agency | Records | Deployed | High-impact | Completeness |
|---|---|---|---|---|
| HHS · Department of Health and Human Services | 447 | 167 | 2 | 78% |
| NASA · National Aeronautics and Space Administration | 425 | 26 | 1 | 63% |
| VA · Department of Veterans Affairs | 367 | 138 | 215 | 57% |
| DOE · Department of Energy | 253 | 67 | 27 | 79% |
| DOJ · Department of Justice | 239 | 88 | 41 | 75% |
| DOC · Department of Commerce | 223 | 0 | 0 | 1% |
| DOI · Department of the Interior | 210 | 29 | 0 | 90% |
| DHS · Department of Homeland Security | 199 | 72 | 16 | 68% |
| USDA · Department of Agriculture | 161 | 38 | 3 | 77% |
| TREAS · Department of the Treasury | 127 | 25 | 4 | 71% |
| DOT · Department of Transportation | 70 | 16 | 0 | 70% |
| SEC · Securities and Exchange Commission | 60 | 20 | 0 | 66% |
| TVA · Tennessee Valley Authority | 59 | 0 | 0 | 17% |
| STATE · Department of State | 58 | 17 | 1 | 60% |
| ED · Department of Education | 56 | 0 | 0 | 17% |
| FDIC · Federal Deposit Insurance Corporation | 50 | 16 | 1 | 62% |
| GSA · General Services Administration | 49 | 17 | 0 | 33% |
| DOL · Department of Labor | 40 | 20 | 0 | 71% |
| FRB · Federal Reserve Board | 38 | 16 | 0 | 84% |
| EPA · Environmental Protection Agency | 27 | 5 | 1 | 66% |
High-impact designation
| Designation | Records |
|---|---|
| High-impact | 317 |
| Presumed high-impact, determined not | 109 |
| Not high-impact | 2468 |
| Not reported | 426 |
Governance evidence: use cases designated high-impact
| Practice | In place | In progress | Not in place | N/A or precluded | Other | No answer | In scope |
|---|---|---|---|---|---|---|---|
| Pre-deployment testing | 0 | 0 | 0 | 0 | 0 | 317 | 317 |
| AI impact assessment | 0 | 0 | 0 | 0 | 0 | 317 | 317 |
| Independent review | 0 | 0 | 0 | 0 | 0 | 317 | 317 |
| Ongoing monitoring | 0 | 0 | 0 | 0 | 0 | 317 | 317 |
| Operator training | 0 | 0 | 0 | 0 | 0 | 317 | 317 |
| Fail-safe | 0 | 0 | 0 | 0 | 0 | 317 | 317 |
| Appeal process | 0 | 0 | 0 | 0 | 0 | 317 | 317 |
| User and public consultation | 0 | 0 | 0 | 0 | 0 | 317 | 317 |
"In place" means the agency reported the practice as established. "No answer" means the field was blank; it says nothing about whether the practice exists.
Method
All figures are computed directly from the normalised inventory records. Category labels are normalised from the source answer options; original values are kept on each record. Governance answers are classified as in place, in progress, not in place, not applicable, precluded, waived, other or no answer, and blanks are never read as "no". See the Methodology page for full rules.
Limitations
- Figures are what agencies reported in the 2025 inventory. They are not verified, and registration of use cases is not complete for every agency.
- A blank answer is missing evidence, not evidence that a practice or safeguard is absent.
- Counts are use-case records. A larger portfolio does not indicate better performance or greater AI maturity, and agencies differ in how they split systems into use cases.
- The inventory has no budget, cost, accuracy or outcome data. None is shown or inferred here.
- This is a disclosure analysis. It is not an audit, a compliance determination or an official government assessment.
Source records
First 40 of 3,320 records, deployed first. The full list is available as CSV from the explorer with the same filters.
Prepared with Federal AI Intelligence by Irakli Petriashvili. Source: https://github.com/ombegov/2025-Federal-Agency-AI-Use-Case-Inventory