The Department of Homeland Security recently published its AI Use Case Inventory, which provides new insight into how DHS agencies, including U.S. Citizenship and Immigration Services (USCIS), are using artificial intelligence (AI) across immigration operations. Several USCIS systems are identified, covering areas such as fraud detection, document processing, and compliance monitoring.
For HR professionals, people leaders, and global mobility teams, the inventory offers a useful window into how the agency is evolving its operations and where employer compliance practices may carry increased significance.
DHS states that the systems listed in the inventory are decision-support tools, not adjudication systems. According to DHS, these tools are designed to identify patterns in data, organize and classify evidence, and assist investigators and analysts. Human officers continue to make all final immigration benefit determinations.
However, some practitioners have reported receiving Requests for Evidence (RFE) that contain language or structure consistent with AI-assisted drafting. This has not been confirmed by USCIS, and there is currently no public evidence that AI is being used to generate adjudication decisions or official correspondence. This observation remains anecdotal and unverified.
One system identified in the inventory is the Text Analytics Data Science Sentence Similarity Model (DHS-130), used by the Refugee, Asylum, and International Operations Directorate.
The tool analyzes digitized narrative text in immigration applications to identify similar language patterns across filings. The goal is to help officers identify patterns that may indicate potential fraud, national security concerns, or public safety risks. The system does not make determinations, but highlights patterns that officers may review during adjudications or investigations.
At present, the inventory indicates this system is used in humanitarian programs rather than employment-based adjudications. There is currently no indication that USCIS is using this model in employment-based cases. That said, the underlying technology illustrates how advances in text analytics could enable the agency to analyze patterns across high volumes of filings in the future. In an employment-based context, that could theoretically include narrative materials such as employer support letters, expert opinion letters, or job descriptions. This is a forward-looking observation, not a current practice.
Another system, the Automated Realtime Global Organization Specialist (ARGOS) for E-Verify Company Registration (DHS-181), analyzes publicly available data about companies registering for E-Verify.
The system searches open-source datasets and applies text analysis to identify risk signals associated with a company registration. Information gathered is ranked and scored to help analysts evaluate whether a registration may warrant additional review. If potential fraud indicators are identified, analysts may refer the case to the Fraud Detection and National Security (FDNS) Directorate for further investigation.
FDNS is the USCIS directorate responsible for detecting immigration fraud, conducting site visits, and coordinating investigative efforts with other federal agencies. FDNS officers conduct worksite inspections, administrative investigations, and verification of employer and employee information. By helping analysts identify potential risk signals earlier, tools like ARGOS may assist USCIS in prioritizing which employer registrations warrant further compliance review.
While the DHS inventory does not specifically reference Project Firewall, employers should be aware of it given its direct relevance to this topic.
Project Firewall is a DOL enforcement initiative launched on September 19, 2025, by the U.S. Department of Labor — not USCIS. It is focused on identifying and investigating employers that may be violating H-1B program requirements. Under Project Firewall, the Secretary of Labor personally certifies the initiation of investigations where there is reasonable cause to believe an employer is out of compliance. The DOL coordinates with USCIS, the Department of Justice, and the Equal Employment Opportunity Commission to share information and refer cases.
The functionality described in AI tools such as ARGOS — analyzing employer information and identifying risk indicators — is consistent with the broader enforcement environment Project Firewall represents. Together, AI-assisted screening and expanded enforcement coordination suggest that employer compliance practices are subject to greater scrutiny than in prior years.
Several systems in the inventory focus on how immigration filings are processed and organized.
The ELIS Evidence Classifier Machine Learning Tagging Solution (DHS-16) automatically tags and categorizes scanned evidence uploaded to USCIS systems, making it easier for adjudicators to locate specific documents. According to DHS documentation, this tool saved approximately 24 million page scrolls between September 2021 and May 2022, and nearly doubled the rate of cases adjudicated within 30 days during that period.
The Intelligent Document Processing for Form Digitization (DHS-2385) uses a learning model to identify and separate uploaded evidence into component documents such as passports, marriage certificates, bank statements, and immigration forms.
The PDF Intake system for myUSCIS (DHS-2598) extracts data directly from uploaded PDF forms, reducing the need for manual data entry.
These systems are designed to reduce manual processing and improve visibility into the information contained in filings.
Historically, immigration petitions have been reviewed largely as individual filings. As USCIS digitizes more filings and deploys analytical tools, the agency may increasingly be able to analyze patterns across employers, industries, job descriptions, filing trends, and supporting evidence over time.
Legal eligibility standards remain unchanged. These tools do not alter what qualifies a petition for approval. However, they may enable USCIS to identify patterns that warrant closer scrutiny across higher volumes of filings than manual review allows.
These developments reinforce the importance of strong immigration compliance practices. As USCIS deploys tools that analyze patterns across filings, consistency and accuracy in documentation take on added significance.
Considerations worth reviewing include ensuring that job descriptions accurately reflect the actual role, maintaining consistency between immigration filings and publicly available company information, avoiding excessive reliance on templated documentation across petitions, and maintaining well-documented immigration processes that can withstand scrutiny.
None of these are new obligations. They are established best practices, and the shift toward more systematic, pattern-based analysis by USCIS is a reason to take them seriously.
Alma helps companies build and manage employment-based immigration programs with the clarity and structure they require. Learn how we can support your team.
