Short answer: DOGE used, or planned to use, AI systems to search and analyze federal contracts, grants, programs, employee spending and other records for possible cuts. Public evidence does not show that DOGE trained a new general-purpose foundation model on government data. “Training an AI” is therefore an imprecise description of a broader program of data ingestion, search, classification and decision support.
What DOGE said it was trying to do
DOGE’s public materials describe a wide-ranging effort to identify waste, fraud and improper payments; review contracts, grants, leases and software licenses; consolidate government information; and recommend programmatic or regulatory changes. Its savings dashboard combines cancellations, workforce changes, regulatory actions and other categories rather than reporting one AI-generated total. DOGE’s dashboard and API are the administration’s own source for those claims.
That distinction matters. “AI analyzed spending” can refer to multiple tools and workflows, not necessarily one identifiable product or model.
What evidence exists that AI was used?
Education Department records
On February 6, 2025, The Washington Post reported that DOGE representatives fed Education Department information into AI software to examine programs and spending. The reported material included internal financial information and personally identifiable information relating to grant administrators. The report did not publicly document the model, its configuration, retention settings or accuracy.
#1 Best Overall
- 40 Color-coded tabs: Highlight the most important sections with over 40 colored tabs for the Applied Behavior Analysis Cooper 3rd edition. The colors match the part for easy reference.
- Find Sections Easily and Efficiently: Our color-coded tabs have large font and are printed on both sides so you can easily navigate the Applied Behavior Analysis Cooper 3rd edition.
- Includes Alignment Card for Perfectly Aligned Tabs: Our tabs are easy to install in a perfect alignment using our tabs alignment system. Each tab includes the location and page number for super easy installation.
- Repositionable: If you misalign the tab no problem! The tabs are repositionable but also once they are folded, stick securely so navigating the ABA is easy and efficient.
- Blank Tabs Included: Additionally we include blank tabs so you can highlight anything specific to your needs.
A proposed central contract repository
The same reporting described a plan associated with GSA technology officials to place federal contract information in a central location for AI analysis. That was a reported plan, not proof that a complete nationwide system was deployed.
Later use of Grok
On May 23, 2025, Reuters reported that DOGE was expanding use of Elon Musk’s Grok chatbot for federal data analysis. Reuters said it could not establish what data had been supplied or how the system was configured. That uncertainty is important when assessing both privacy and conflicts of interest.
A separate regulation-focused tool
In July 2025, The Washington Post described a “DOGE AI Deregulation Decision Tool” intended to analyze approximately 200,000 federal regulations. Regulatory analysis is related to DOGE’s automation strategy, but it is not evidence of an AI trained to analyze spending.
Rank #2
- The 2024 ERG guide helps satisfy 49 CFR 172.602 DOT requirement. This requirement states that hazmat shipments be accompanied by emergency response info.
- Pocketbook aids in emergency preparedness, planning, and training with ERGs numerically indexed and color-coded to help emergency responders find vital information fast.
- 2024 Updates: The Pipeline and Hazardous Materials Safety Administration (PHMSA) released a comprehensive summary of updates. Most significantly a QR code on the back cover that provides access to critical incident reporting information.
- Other changes for 2024 have been made to continue to provide the most accurate emergency response information to help all front-line persons and all first responders stay safe during transportation emergencies.
- Specifications: 4" x 5 1/2" Pocketbook Size, English, Spiralbound. Copyright 2024.
“Training” is not the same as analyzing records
Several technically different activities are often compressed into the word training:
Quick wins for a faster PC:
Repair Windows errors before they cause bigger problemsFix Now →Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →Clear out junk files and repair common Windows errorsFree Scan →| Activity | What it means | What is publicly established about DOGE |
|---|---|---|
| Prompting or querying | An existing model receives questions and selected records and returns an answer. | Reported as a possibility; the specific systems and prompts are not documented publicly. |
| Retrieval-augmented analysis | A search index retrieves private documents or database rows for an existing model to summarize or classify. | Consistent with reported plans, but no complete architecture has been published. |
| Fine-tuning | An existing model’s parameters are adjusted using a specialized training set. | No public technical documentation establishes that DOGE fine-tuned a model. |
| Pretraining a new model | A foundation model is built from a very large corpus and trained from the beginning. | No public evidence identifies a DOGE foundation model, architecture, corpus or developer. |
| Conventional analytics | Rules, database queries, statistical tests or automation flag records without model training. | Could be part of the program even when described publicly as AI. |
Accordingly, the strongest defensible description is “AI-assisted government-data analysis,” not “DOGE trained an AI on government spending.”
What kinds of data were at stake?
| Data category | Examples and principal risk |
|---|---|
| Public procurement records | Contracts and awards; errors can cause misclassification or duplicate counting. |
| Internal financial records | Agency spending and obligations; unauthorized disclosure or context loss. |
| Taxpayer information | Returns and account data subject to strict privacy and access rules. AP reported DOGE sought access to sensitive IRS data. |
| Student-aid and grant records | Personally identifiable and financial information, including the Education Department material reported in February 2025. |
| Personnel and communications | Potential surveillance, retaliation or disclosure concerns. |
| Contractor information | Trade secrets, nonpublic pricing and protected procurement information. The Washington Post reported access to such records across agencies. |
| Regulations and guidance | Legal misinterpretation when text is treated as a simple classification problem. |
These categories are not interchangeable. A public contract award does not carry the same legal or security consequences as a tax return, personnel file or contractor trade secret.
Rank #3
- EASY TO MANAGE - Use this income & expense log book to record your income and expenses each day.Keep your budget in balance, and develop good bookkeeping habits to meet your financial goals
- ACCOUNTING FOR THE WHOLE YEAR - This income and expense tracker is undated and is used to lasts a whole year.The keeping log has 1 page Year Overview, 53 weekly spreads, 2 pages annual summary, 10 notes pages, to track weekly and yearly income & expenses
- HIGH QUALITY - The accounting bookkeeping tracking ledger log book is used to high quality 100gsm pure white paper, teal elastic band and a back pocket for extra space. Make sure you have enough space for all financial activities
- UNIQUE DESIGN & A4 SIZE - Income and expense log book is spiral bound design, size of 8" x 10.5". Just the perfectly size to fit in your backpack, purse or laptop case. Without taking up your space and always helping you keep track of your small business
- THE PERFECT GIFT - Income & expense notebook as gift for woman & man. Use it to track your week-to-week progress, make efficient adjustments whenever needed
How a spending-analysis system could work
The following is an explanatory architecture, not a confirmation of DOGE’s exact implementation:
- Ingest data: import contracts, grants, invoices, leases, payment records and agency databases.
- Normalize records: standardize vendor names, agency codes, dates, amounts and contract identifiers.
- Match entities: connect parent companies, subsidiaries, subcontractors, programs and offices.
- Retrieve evidence: search documents and structured data in response to analyst questions.
- Classify spending: label records by program, authority, vendor, policy category or possible redundancy.
- Detect anomalies: flag duplicate payments, dormant licenses, unusual prices, expired contracts or inconsistent records.
- Review findings: contracting officers, auditors, lawyers and program experts verify each flag.
- Take action: recommend cancellation, renegotiation, consolidation, investigation or no change.
- Preserve an audit trail: retain source records, model and prompt versions, rules, reviewers and final decisions.
An AI flag is not proof of waste. A system may spot apparently redundant contracts while missing statutory duties, option years, cancellation costs, safety requirements or the difference between a contract ceiling and money actually spent.
Why “spending” is difficult to measure
- Ceiling versus obligations: a contract’s maximum potential value is not necessarily money committed.
- Obligations versus outlays: an obligation is a commitment; an outlay is money actually disbursed.
- Gross versus net savings: termination fees, replacement contracts and remediation can reduce or eliminate a claimed saving.
- One-time versus recurring savings: canceling a subscription may recur, while canceling a project may produce a single avoided expense and new costs elsewhere.
- Apparent duplication: similar contracts may serve different agencies, security environments or locations.
- Social and legal value: a low-dollar program can have substantial public or statutory importance.
DOGE says its savings calculation can reflect the gap between a contract’s total value and the amount currently obligated, and notes that Federal Procurement Data System postings can lag by up to a month. That methodology is not the same as independently verified cash savings.
Rank #4
- 4 Pack Spending Account Tracker Notebooks
How reliable are the published savings figures?
DOGE’s website listed $215 billion in estimated savings as of January 1, 2026. The site also said its public receipts represented approximately 30 percent of total contract, grant and lease cancellations. Those are DOGE estimates, not an independent audit of budgetary savings.
The relevant questions are:
- Is the figure based on ceilings, obligations, outlays or projected future reductions?
- Were cancellation payments and replacement costs deducted?
- Were duplicate actions removed?
- Can another auditor reproduce each calculation from the receipts and source records?
- Did the reduction persist after implementation?
AP later reported that federal auditors found some DOGE savings claims incorrect or unsupported. It is therefore more accurate to write “DOGE estimated” or “DOGE claimed” than “AI found $215 billion in waste.”
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Privacy, security and conflict-of-interest risks
- Personally identifiable information may be entered into a commercial model or retained in logs.
- Combining agency databases creates a high-value target and magnifies the consequences of one breach.
- Temporary or politically appointed personnel may receive access without the same operational experience as career specialists.
- Malicious text in a document can attempt prompt injection or manipulate automated classification.
- Hallucinated accusations of fraud or waste can influence layoffs, cancellations or investigations.
- Trade secrets and nonpublic contractor information may leak across agency boundaries.
- Unclear deletion, retention and audit policies make later accountability difficult.
- Use of a Musk-affiliated company’s AI in a Musk-led initiative raises a conflict question even when no violation has been established.
Access, processing, disclosure, retention and a confirmed breach are different events. Public reporting establishes concerns and reported access, but does not by itself prove that every record was disclosed or used to train a vendor’s general-purpose model.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
Best Value
Legal and oversight questions
The public record leaves unresolved whether each system and data source had the required authorization, whether Privacy Act, records, procurement and cybersecurity rules were followed, and whether access logs and data-use agreements were maintained. It also raises questions about training, financial disclosures, conflicts, human approval and avenues for agencies or contractors to correct errors.
A March 2025 congressional document asked agencies about DOGE’s AI use and data used to develop or train an algorithm. A New York judge later relaxed restrictions on DOGE access to sensitive Treasury information subject to conditions including required training and financial-disclosure requirements for a DOGE worker. AP reported on that ruling.
GAO’s broader federal work identifies recurring procurement problems: agencies may struggle to evaluate vendor proposals, understand AI-related costs and capture lessons from purchases. See GAO’s acquisition report and its report on federal AI uses and risks. A separate GAO review found no evidence that DOGE detailees accessed NLRB IT systems during the reviewed April 16–July 25, 2025 agreement period; that finding concerns one agency and period, not all federal access. GAO’s NLRB review explains the scope.
What remains unknown
- Which models were used by each agency and whether Grok was used in the Education Department case.
- Whether any government records were used for pretraining or fine-tuning a commercial model.
- What retention, deletion and isolation policies applied.
- How often human reviewers overturned AI recommendations.
- Whether AI outputs directly caused particular cancellations.
- Whether savings estimates were independently audited.
- Whether DOGE used one platform or multiple agency-specific systems.
What a credible evaluation would require
- Complete, current and deduplicated source data.
- Documented legal authority for every dataset and user.
- Disclosure of model, retrieval corpus, prompts, rules and versions.
- Measured false-positive and false-negative rates.
- Named human officials accountable for each decision.
- Reproducible calculations separating ceilings, obligations, outlays and net savings.
- Conflict controls for vendors and officials with relevant private interests.
- Security controls that prevent commercial model training or unauthorized retention.
- An appeal and correction process for agencies, contractors and beneficiaries.
- Public audit records showing whether projected savings actually materialized.
Bottom line
DOGE’s effort is best described as AI-assisted government-data mining and decision support. Reporting supports the use or planned use of AI to inspect federal records, including sensitive information, but does not publicly demonstrate a newly trained DOGE foundation model. The credibility of any resulting savings depends less on the presence of a chatbot than on lawful access, data quality, transparent accounting, independent auditing, secure system design and accountable human review.
Do these 3 things before closing this tab:
1Fix the driver behind crashes, sound loss and screen glitches2Repair Windows errors before they cause bigger problems3Scan for outdated or missing drivers - takes under a minuteQuick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
