On Aug. 4, the U.S. Department of Justice announced a combined $3.2 million settlement with OpenAI OpCo LLC and its subsidiary Statsig Inc. resolving allegations that the companies discriminated against U.S. workers during recruitment for positions connected to the permanent labor certification process, highlighting renewed federal scrutiny of PERM recruitment.

A separate February DOJ settlement with Elegant Enterprise-Wide Solutions Inc. further demonstrates that employers remain responsible when artificial intelligence produces or influences potentially discriminatory recruitment practices.

Whether businesses are AI providers or AI users, they need to pay attention to the increasing government scrutiny of this cutting-edge technology. In particular, as employers increasingly use AI to perform employment functions — such as drafting job postings, screening applicants, and managing growing volumes of automated or AI-generated applications — companies should evaluate whether their recruitment technology is consistent with immigration and employment discrimination laws at the federal, state and local levels.

The OpenAI Settlement

According to the DOJ, OpenAI and Statsig did not post job openings for PERM positions on their external career websites even though other job openings were routinely posted there. Both companies also required applicants for PERM positions to mail paper applications, despite accepting electronic applications for other positions.

The DOJ further alleged that other PERM-related recruitment practices discouraged applications, including advertising PERM job openings on the radio late at night.

Under the settlement, OpenAI agreed to pay $1.2 million in civil penalties, establish a $2 million back pay fund, revise its PERM recruitment processes — including by posting PERM positions on its public career website and accepting electronic applications — and comply with government monitoring and reporting requirements.

Notably, fewer than 10 PERM positions were at issue, so the steep settlement underscores the DOJ's intention to seek maximum civil penalties even in cases involving a limited number of positions.

In light of this settlement and the broader enforcement environment, employers of all sizes should proactively evaluate their current PERM recruitment processes and make any necessary adjustments to ensure compliance with applicable legal requirements and trends.

This was the 13th settlement since the DOJ relaunched its Protecting U.S. Workers Initiative in 2025. Originally launched during President Donald Trump's first term, the initiative targets, investigates and brings enforcement actions against employers that allegedly intentionally discriminate against U.S. workers due to a preference for temporary visa workers. The pace of these settlements underscores the DOJ's focus on these issues.

An Established and Expanding Enforcement Pattern

The OpenAI settlement follows several significant PERM-related enforcement matters.

For instance, in 2023, Apple agreed to pay up to $25 million to resolve the DOJ's allegations that it discriminated against U.S. workers by not posting PERM positions on its external career website, requiring paper applications and failing to consider certain electronically submitted applications for PERM-related job openings.

Additionally, in April, the DOJ filed a lawsuit against software firm Cloudera Inc., alleging that the company discriminated against U.S. workers by creating a separate recruitment process for PERM positions and instructing applicants to use an email address that could not receive external applications.

While litigation remains ongoing, in May, the U.S. Department of Labor suspended processing of all pending Cloudera PERM applications for 180 days. The suspension highlights the DOL's willingness to halt employer immigration benefits when PERM recruitment integrity is in question, even when there has not yet been a judgment or settlement, underscoring the serious business consequences of perceived noncompliant recruitment practices.

These matters, along with other DOJ enforcement actions, demonstrate that regulators may compare an employer's PERM recruitment methods with its ordinary hiring practices. Separate application channels, less accessible postings or additional procedural barriers may attract scrutiny even when the employer complies with the minimum advertising steps required under the PERM regulations.

Finally, the private plaintiff class action bar has followed the government's lead in seeking to target and pursue enforcement measures against companies that discriminate against U.S. workers in favor of foreign visa workers. For instance, in Devorin v. Tata Consultancy Services Ltd. in 2024, a former employee filed a class action against Tata in the U.S. District Court for the Southern District of Florida, alleging it terminated American workers at its U.S. locations to replace them with foreign workers on temporary visas.

Why AI Does Not Shift Responsibility

Two complementary legal regimes apply. Under Title 20 of the Code of Federal Regulations, Section 656.10(c)(8)-(9), a PERM position must be clearly open to any U.S. worker, and U.S. workers may be rejected only for lawful, job-related reasons. Separately, Title 8 of the U.S. Code, Section 1324b, prohibits certain discrimination based on citizenship status and national origin in hiring and recruitment.

Accordingly, compliance with the DOL's minimum PERM advertising requirements does not necessarily eliminate potential exposure under the Immigration and Nationality Act's antidiscrimination provisions.

Moreover, employers are increasingly turning to AI-powered tools to streamline hiring and recruitment processes. These technologies can automate job postings, screen applications and generate candidate communications at scale. However, adopting AI tools in recruitment introduces potential legal risks.

When AI systems produce content or make screening decisions that reflect unauthorized preferences for citizenship status or national origin — whether through algorithmic bias, flawed training data or inadequate human oversight — employers may face liability under the INA, regardless of whether the discriminatory output was intentional.

The DOJ's February settlement with Elegant Enterprise illustrates these risks. In that matter, the DOJ alleged that job advertisements generated by an AI tool contained unauthorized restrictions based on visa status, limiting consideration to applicants holding H-1B, optional practical training or H-4 work authorization.

The settlement underscores that employers remain responsible for all hiring-related communications — regardless of whether they are drafted by human personnel or generated by AI.

Accordingly, employers should implement meaningful review processes to ensure that AI tools do not incorporate impermissible restrictions into their recruitment and hiring activities, as even unintentional discriminatory language can expose employers to civil penalties, government oversight and reputational harm.

Screening AI-Generated and Automated Applications

Employers are also facing a different AI-related problem within recruitment and hiring: rapidly increasing numbers of applications prepared or submitted through generative AI and automated tools. These applications may contain standardized language, duplicate submissions, exaggerated qualifications or information that is difficult to verify.

The volume of such submissions also creates significant administrative burdens, including difficulty distinguishing authentic applications from automated responses, increased processing demands on human resources personnel, and challenges in maintaining complete and accurate recruitment records required for PERM audit purposes.

However, these administrative challenges do not modify the PERM standard or permit any reduction in the rigor of application review. Using AI to prepare or submit an application does not, by itself, relieve an employer of its obligation to assess whether a U.S. worker applicant is potentially qualified under the applicable PERM standard.

For instance, an applicant should not be rejected merely because a résumé appears to have been prepared using AI or submitted through an automated process. The employer must still determine whether each applicant satisfies the specific stated minimum requirements for the PERM position or whether they could acquire a missing skill through a reasonable period of on-the-job training.

In certain cases, AI-enabled screening may create additional risk when a system:

• Rejects applicants based on keywords, formatting, employment gaps or inferred characteristics, rather than the stated PERM requirements;
• Applies qualifications that were not disclosed in the PERM advertisements;
• Treats applicants for PERM positions differently from applicants for comparable non-PERM positions;
• Screens applicants based on citizenship or visa status terminology; or
• Does not retain sufficient information to reconstruct and explain an individual rejection.

These AI-related screening risks are not unique to the PERM context. The U.S. Equal Employment Opportunity Commission issued guidance in 2022 and 2023 reinforcing that existing federal employment discrimination laws apply even when an automated system participates in hiring decisions. Although the EEOC later withdrew that guidance as potentially out of date, federal employment laws remain fully applicable to AI-driven hiring.

For example, employers may face disparate impact liability when AI tools utilized in hiring disproportionately exclude or disadvantage a protected group, even if the company had no intent to discriminate. Likewise, the Americans with Disabilities Act imposes liability on employers with AI applicant review programs that screen out candidates based on disability-related characteristics, such as video analysis or timed responses that inadvertently screen out qualified individuals with disabilities.

Employers must also ensure that their automated systems can accommodate applicants who request reasonable accommodations during the application process. Many state fair employment laws provide broader protections than their federal counterparts, covering additional protected classes or applying to smaller employers. AI-driven screening decisions that produce disparate impact may trigger liability under these statutes, independent of any federal claim.

Recent litigation further underscores the breadth of employment discrimination exposure arising from AI screening tools. In Harper v. SiriusXM Radio LLC, a class action filed in the U.S. District Court for the Eastern District of Michigan last year alleged that AI-powered applicant screening tools discriminate against African American applicants by relying on data points that serve as proxies for race, asserting claims under both Title VII and Title 42 of the U.S. Code, Section 1981.

In 2023, in Mobley v. Workday Inc., plaintiffs in the U.S. District Court for the Northern District of California challenged Workday's use of AI to facilitate employment searches, alleging the outputs resulted in disparate impact based on race, sex, age and disability. Further, a conditional Age Discrimination in Employment Act collective action has attracted thousands of opt-ins.

For employers conducting PERM recruitment, this means that AI screening tools must comply not only with PERM and INA requirements, but also with broader federal antidiscrimination standards, including Title VII, the ADEA, the ADA, Section 1981 and analogous state fair employment laws — creating multiple, overlapping compliance obligations that require careful coordination.

These cases remain pending, and the allegations have not been finally adjudicated. Nevertheless, they illustrate potential exposure for both employers and technology providers when automated tools materially participate in employment decisions.

Navigating Overlapping Compliance Obligations

Employers that sponsor workers through the PERM labor certification process face a challenging compliance landscape. The DOL's PERM regulations impose strict procedural requirements governing job advertisements, recruitment steps and documentation, all designed to test the labor market and demonstrate that no qualified U.S. worker is available for the position.

At the same time, the DOJ's Immigrant and Employee Rights Section enforces the antidiscrimination provisions of the INA, which prohibit recruitment and hiring discrimination based on citizenship status and national origin. This creates an inherent tension: Certain recruitment practices that are undertaken to satisfy PERM requirements — such as specific language in job postings, narrow applicant screening criteria or separate application procedures — may inadvertently trigger DOJ scrutiny if they appear to discourage or exclude U.S. workers.

The increasing use of AI tools in hiring adds another layer of complexity, as automated systems may inadvertently produce discriminatory outputs or screening decisions that expose employers to liability under both regulatory frameworks. Given the complexity of these overlapping obligations, employers should design PERM recruitment processes that satisfy the DOL's procedural requirements, while remaining fully compliant with the DOJ's antidiscrimination standards.

What This Means for Employers

The OpenAI and Elegant Enterprise settlements offer critical lessons for employers navigating AI-driven recruitment. First, even sophisticated technology companies face significant liability exposure when PERM recruitment processes deviate from standard hiring practices. Second, employers bear responsibility for all hiring-related communications — including those generated by AI tools — and cannot delegate compliance obligations to technology vendors.

All the aforementioned cases underscore that businesses functioning as AI providers, AI users or both must proactively evaluate whether their recruitment technology and practices align with federal antidiscrimination provisions, DOL PERM requirements, and applicable state and local employment laws.

As AI tools become increasingly foundational to core employment functions — from drafting job postings and screening applications to managing the growing volume of automated and AI-generated submissions — companies must establish robust compliance frameworks or risk becoming subject to aggressive enforcement consequences.

Thus, employers using AI or automated tools in the PERM recruitment context should consider taking the following steps.

1. Use substantially comparable application methods for PERM and ordinary recruitment.

2. Post PERM openings on the company's public career site when comparable positions are ordinarily posted there.

3. Require meaningful human review before an applicant is rejected through automated screening.

4. Configure screening criteria to match the stated PERM requirements exactly.

5. Review AI-generated advertisements for unauthorized citizenship or visa status restrictions.

6. Document individualized, lawful and job-related reasons for each rejection.

7. Preserve applications, screening results, system configurations, communications and audit trails.

8. Continue to thoroughly evaluate all eligible applications to determine if a candidate is potentially qualified, even if the application appears to be automated or assisted by AI.

9. Audit recruitment vendors and applicant-tracking systems, rather than relying solely on vendor assurances.

10. Determine whether state or local bias audit, notice, accommodation or disclosure requirements apply.

11. Coordinate immigration counsel, employment counsel, human resources, recruiters and technology teams before beginning recruitment.

The recent enforcement actions demonstrate that PERM recruitment cannot be treated as a technical advertising exercise. As AI becomes more involved on both sides of the application process, employers must be prepared to show that their recruitment was accessible, nondiscriminatory, conducted in good faith and supported by a reliable record of individualized decision-making.

Reprinted with permission of Law360.