The Evolving Role of AI in the Workplace: Legal and Practical Considerations for Employers

Artificial intelligence (AI) has moved quickly from a theoretical consideration to a practical reality for employers, with tangible implications for workforce management. With offices in both the US and the UK, we see employers across both jurisdictions confronting many of the same practical questions around workplace AI, even as the applicable legal and regulatory frameworks continue to evolve in different ways. Findings from the U.S. Census Bureau’s March 2026 Household Trends and Outlook Pulse Survey indicate that approximately 56% of US workers have used AI for at least one of 11 workplace tasks, and among those workers, approximately 24% reported using AI every day during the preceding week. Similarly, the UK government research published in January 2026 indicated that 31% of UK employers are already using or deploying AI in their day-to-day operations. Other recent studies suggest that organizations are increasingly integrating AI into human resources (HR) functions, particularly in recruiting, learning and development, and performance management.

Against that backdrop, this article summarizes some of the key AI-related questions arising for employers in the US and the UK and highlights legal and workforce governance issues employers are navigating as technology and regulation continue to evolve. This article is not a comprehensive review of all HR-related issues in this space. Importantly, these issues need not be viewed as barriers to adoption. By embedding practical and appropriate safeguards, governance structures, and compliance measures into their AI strategies, employers can harness the advantages of AI while reducing legal and operational exposure.

Hiring and Recruitment

Employers are increasingly using AI in their hiring and recruitment processes, including to identify and target candidates, sift applicants, screen résumés, and track application progress.

While the efficiencies are clear, embedding AI in hiring processes or otherwise relying on AI agents to help identify and sift candidates without sufficient safeguards carries significant risks. Most prominently, bias in AI algorithms and/or source data can give rise to discrimination claims. For example, in the UK, if AI deployed by an employer in their hiring process selects or prioritizes (or alternatively, rejects or deprioritizes) candidates as a result of a protected characteristic set out in the UK Equality Act 2010, the employer can be liable for discrimination claims. Similar risks arise in the US, where AI-assisted recruiting and hiring practices may expose employers to discrimination claims under federal, state, and local antidiscrimination laws if their use results in unlawful discrimination or disproportionately disadvantages members of protected groups. US employers should also be mindful of accommodation obligations, as certain AI-driven assessments or interview tools may disadvantage qualified applicants with disabilities absent appropriate accommodations.

Outsourcing hiring to third-party vendors does not eliminate these risks. In both the US and the UK, employers generally remain responsible for ensuring that AI-assisted hiring processes comply with applicable employment laws, even when those processes are administered through a third-party vendor. To mitigate these risks, employers should use AI products only for their intended and documented purposes and avoid modifying or deploying tools in ways not contemplated by the vendor. Employers should also conduct appropriate diligence on any third-party provider, including understanding how their tool functions, what data it uses, what safeguards and testing have been implemented, and how the vendor addresses applicable legal requirements. Vendor agreements should be carefully reviewed to understand responsibility for compliance, audit rights, data usage restrictions, indemnification provisions, and limitations of liability. Employers may also wish to conduct periodic internal testing or audits to confirm that AI-assisted hiring processes continue to operate as intended and remain compliant with applicable legal requirements.

In addition, in the US, several state and local laws expressly regulate the use of automated decision-making tools in employment contexts, including tools supplied by external vendors. Similarly, the use of AI in recruitment can also raise data privacy concerns in the UK if automated decision-making is used without employers’ knowledge and appropriate safeguards in place. In the UK, the Information Commissioner’s Office recently found that a number of employers that “considered it unlikely” they were using automated decision-making in their recruitment were, in fact, doing so (and almost 70% of organizations surveyed anticipate increasing the use of AI and automation in recruitment processes going forward). In practice, many third-party vendors often cap their liability in a way that can limit meaningful recourse for employers, making it essential to interrogate how exactly AI is being used in these situations and what the liability provisions in the applicable services agreement say. US employers should also consider applicable federal, state, and local privacy, consumer protection, and AI-specific laws and ensure that AI-assisted recruiting tools do not collect, process, or generate information that could result in prohibited inquiries, such as inquiries regarding an applicant’s disability or medical condition under the Americans with Disabilities Act (ADA).

Employers should also be mindful of applicable notice, transparency, and recordkeeping requirements, which extend beyond hiring and recruiting to other employment-related decisions. In the US, a growing number of state and local laws, such as those in Illinois, Colorado, California, Connecticut, and New York City, impose or will impose varying obligations — including notice of AI tool use, disclosure of adverse decisions or review rights, audits or assessments, and record retention — when AI is used to make or influence employment-related decisions. A one-size-fits-all approach across multiple jurisdictions carries compliance risk, as requirements vary significantly by country, state, and city. Employers hiring across international lines are strongly encouraged to consult employment counsel to build an AI hiring strategy that complies with applicable laws.

Performance Management and Workforce Decisions

AI is also increasingly being used to monitor employee performance, evaluate productivity, and identify employees for redundancy or layoff. These applications carry the same bias, transparency, and data privacy risks previously noted and introduce additional legal exposure. In the UK, employers must comply with UK General Data Protection Regulation obligations on automated decision-making, consider whether the AI-driven selection criteria result in indirect discrimination under the Equality Act 2010, and consider collective consultation requirements when AI tools inform redundancy selections. Similar considerations apply in the US, where existing employment laws generally apply regardless of whether decisions are made by human managers, AI systems, or a combination of both. Employers should carefully assess whether AI-assisted processes used for performance management, compensation, promotion, discipline, termination, or workforce reductions could create unintended discriminatory outcomes, disability accommodation concerns, privacy issues, or other compliance risks. They should also remain mindful of evolving state and local AI laws, which increasingly impose requirements relating to notice, transparency, documentation, auditing, record retention, and, in some cases, human review rights.

Before deploying AI for these purposes, employers should work with counsel to establish appropriate governance measures, including meaningful human oversight of significant employment decisions, documentation of decision-making processes, and periodic review of AI-assisted workflows to confirm they are operating as intended.

Human Review and Oversight

We frequently advise employers on strategies for human review and oversight when implementing AI tools into HR processes. Employers should embed meaningful human review at critical decision points in any AI-assisted HR workflow. While AI tools deliver significant efficiencies, human oversight is an important check on AI outputs and can substantially reduce (though not eliminate) the legal and operational risks previously discussed. In fact, the trend toward mandated human oversight is already emerging in the US: Recently enacted California law, effective July 1, 2027, prohibits employers from relying solely on automated decision systems in disciplinary or termination decisions and requires a human to corroborate any system output that the employer primarily relies on, while recently enacted Colorado law, effective January 1, 2027, provides individuals “the right to request meaningful human review and reconsideration” after certain adverse decisions materially influenced by automated decision-making technology are made. Other jurisdictions may soon follow suit.

The appropriate level of review will depend on the nature of the decision, the potential impact on employees or applicants, and applicable legal requirements. In our experience, a risk-based or tiered approach to integrating human oversight of AI-assisted workflows is often the most effective means of managing risk while preserving the value of AI-enabled tools.

Regardless of the approach adopted, employers should document their governance and review procedures, train personnel to exercise independent judgment when evaluating AI-generated outputs, and periodically assess whether those controls are functioning as intended. A human “rubber stamp” on an AI recommendation is unlikely to satisfy legal expectations or provide meaningful protection against claims.

Policies, Training, and Thoughtful Implementation

Employers are increasingly aware that their workforce uses AI — whether through formally integrated tools or employees’ personal applications used without oversight. Without appropriate checks and balances, this can give rise to significant liability. For example:

• AI summary or note-taking tools on calls may produce inaccurate recordings or breach confidentiality and data privacy requirements. In the US, many jurisdictions have wiretapping laws requiring two-party or all-party consent; failure to obtain consent before using AI to record or summarize discussions may violate such laws. Employers should also avoid inadvertently capturing sensitive employee information (such as medical, financial, or performance-related data) subject to heightened privacy protections under applicable laws such as the ADA.
• The use of AI tools and large language models may implicate privilege protections. Employees and employers should be particularly careful and consult counsel before using AI in connection with any litigation, potential claim, or any other sensitive or potentially privileged matter.
• Uploading sensitive business information to large language models using public versions of products may inadvertently disclose confidential information, implicating data privacy laws and trade secret protections or creating cybersecurity liabilities.
• Employees who rely on unverified AI outputs may create additional risk if they deliver inaccurate or hallucination-affected products or services to clients. Employers will likely be vicariously liable for such actions under English law, and US employers may face similar risks under applicable laws.

Such risks do not mean employers should not let their workforce use these tools — but it means they should be thoughtful about how they do it. To start, employers should develop employee-facing AI policies for their workforce, as discussed in Goodwin’s recent article “Every Company Needs an AI Use Policy,” and consider tying breach of such policies to disciplinary action. At a minimum, such policies should clearly explain when AI and recording tools can and cannot be used; when and how employees are expected to obtain consent or provide notice before using an AI or a recording tool; what qualifies as appropriate use (including reminders that the company’s antidiscrimination, anti-harassment, code of conduct, and other policies apply in the AI space); and which AI and recording tools have been approved for use by the organization as well as any other applicable cybersecurity standards and protocols. Of course, such generally applicable policies are in addition to any function-specific protocols, such as HR and IT policies and broader organizational governance protocols. Such policies should remain subject to ongoing review by counsel and employers as laws and standards continue to evolve. Notably, we are also seeing these issues receive increasing attention in transactions, in which buyers are keen to understand a target’s use of AI and associated governance frameworks.

Policies alone are not sufficient, however. Employers should provide onboarding and ongoing training to ensure staff are aware of, and can apply, AI best practices and applicable restrictions. Training can be effectively integrated into existing modules on antidiscrimination and anti-harassment, with specific content for managers, HR, and other stakeholders.

Workforce Restructuring

There are increasing reports of companies undertaking layoffs or redundancies because of, or in anticipation of, roles being replaced by AI. Such exercises have also drawn criticism from employees who view such redundancies as “AI washing,” with companies using AI as an excuse to implement unpopular cost savings. While the fairness of such layoffs in the UK hinges on whether there is a “genuine” redundancy, in our experience, tribunals are generally reluctant to substitute their view on whether an employer had “valid” business reasons for undertaking redundancies. By comparison, in the US, where employment is generally at will in most jurisdictions, such issues largely present a workforce relations issue rather than a new liability concern in and of itself. That said, at least three US jurisdictions have recently taken steps to collect information concerning AI-related workforce reductions. Connecticut (effective October 1, 2026) and California (effective January 1, 2027) have enacted legislation requiring AI-related disclosures in connection with certain Worker Adjustment and Retraining Notification Act (WARN) notices, while New York has added an AI- and automation-related disclosure to its WARN filing system. Other jurisdictions may follow in requiring such disclosures as state legislatures continue to respond to the effects of AI on the workforce. Accordingly, US employers should remain vigilant of any additional requirements for layoffs as such laws continue to evolve. As a result, we expect AI may continue to feature strongly in employee town halls and explanatory documents as a primary reason employers need to make redundancies.

Grievances and Litigation

AI is having a notable impact on employee grievances and litigation, particularly in the UK. We are seeing a rise in employees using AI tools to draft grievances and bring legal claims, generating increasingly complex, overstated, or unsubstantiated allegations that present new challenges for employers. In the UK, this is also affecting Employment Tribunal resources, as unrepresented claimants gain a potentially false sense of confidence in the merits of their claims.

In both the UK and the US, the use of AI in litigation proceedings (by employers and employees) will have implications for legal privilege (as summarized in previous Goodwin articles: “Proceed With Caution: AI Platforms and Privilege” [UK] and “AI Chatbots, Privilege, and Pitfalls: Lessons for Keeping Generative AI Exchanges Out of the Hands of Legal Adversaries” [US]). For example, using AI summary or note-taking tools in otherwise privileged discussions, or uploading potentially privileged information to AI large language models or other AI tools, could subject such privilege to credible challenge in litigation. Similarly, HR professionals should be cognizant of the fact that their own use of AI research functions may be subject to discovery in litigation if such use was not directed by their attorney in reasonable anticipation of litigation. For businesses facing litigation, ensuring appropriate guardrails are in place on the use of AI to protect their legitimate interests (including inadvertent risk of privilege being waived) will be of increasing importance.

We expect the trend of self-represented litigants resorting to AI will continue and have a significant impact on workplace disputes. Particularly in light of the incoming changes in UK employment law (e.g., the reduction in the requisite service employees need to bring unfair dismissal claims, the removal of the compensation cap on unfair dismissal claims, and extension of tribunal time limits), this will likely result in more protracted settlement negotiations between parties, increased litigation, and further delays in resolving employment disputes.

* * *

AI is clearly here to stay and is having a significant impact on the workforce. While the law is yet to catch up in several key aspects, employers are dealing with novel and complex issues on a regular basis. Employers that are proactive in how they manage deploying AI in their businesses and assessing risks presented by AI will be better equipped to deal with issues impacting their workforces. The issues discussed in this article are real, and employers should thoughtfully evaluate the legal, compliance, and workforce implications of AI adoption.

At the same time, these considerations should not deter organizations from embracing technologies that can drive significant efficiencies, innovation, and business value. With appropriate governance, clear policies, employee training, and ongoing oversight, employers can responsibly integrate AI into the workplace while mitigating associated risks.

Please reach out to your Goodwin contact or any of this alert’s authors should you want to discuss any of the matters raised in this briefing in more detail.