This article was co-authored by Labor + Employment group lawyer, Christopher Costain.
As artificial intelligence (AI) continues to play a significant role in how employers streamline operations, simplify workflows, and increase productivity, some employers are turning to highly advanced “AI agents” to take over multi-step, complex tasks that go far beyond the AI chatbots with which employees have become familiar. Now, some employers are taking the next step by using such highly advanced AI agents with human names to complete complex, multi-step assignments—not simply respond to prompts. However, employers should consider the extent to which giving AI agents human names, job titles, and places on an organizational chart may impact their human employees.
An AI agent is an autonomous, goal-driven, digital system that uses a large language model embedded with an organization’s resources, tools, and information to reason, plan, and make decisions without meaningful human oversight. Unlike an AI chatbot that provides a response to a prompt, an AI agent can coordinate actions across business functions and manage an entire decision-making process from start to finish. Since AI agents can tackle more complex tasks, employers have given them email addresses and started referring to them as “colleagues” who have completed onboarding and orientation. That may make use of the technology feel more accessible, but it can also blur the line between a software tool and a human decision-maker.
Although AI agents offer exponential improvements in productivity and efficiency, there are risks that employers should consider. Like all AI software, human engagement, judgment, and accountability should be a routine part of the work performed by any AI agent. Employers should ensure that employees understand that, even though they can execute tasks at scale, ultimately, AI agents cannot be held accountable when things go wrong; that responsibility falls on the employee who deployed the AI agent for the particular task, regardless of the fact that the AI agent has a human name, is referred to as a “coworker,” or maybe even is situated above them on the organizational chart. To that end, employers should also consider where an AI agent falls within the organizational structure, who will be responsible for “supervising” the AI agent, and what that “supervision” will be. Employers should also ensure their AI-related workplace policies are clear that employees, not AI agents, are responsible for producing high-quality work product and for verifying the accuracy of an AI agent’s outputs, regardless of the fact that the AI agent is performing a multi-step task and engaging in high-level functioning across business functions.
Employers should also be cognizant of how an AI agent might impact employee relations and create employment risk. As AI agents become more adept at analyzing complex employment law issues (including, for example, related to job-protected leaves of absence and reasonable accommodations), employees are likely to turn to their AI agent colleagues for answers and guidance based on their unique circumstances. In light of this, employers should consider whether AI agents, who communicate with employees through a company email address and with a corporate signature, making those communications indistinguishable from a human actor, are capable of binding the company to a particular accommodation or leave arrangement in the same manner that a human agent could bind the company. Separately, employers should also review their leave and accommodation-related workplace policies and practices to ensure they are consistent with the law and with the information an AI agent is trained on and may ultimately provide to employees, so that the AI agent’s responses are consistent with those policies and practices.
Additionally, employers might consider the impact that AI agents will have on employee morale. Will employees like working with “AI colleagues”? An AI agent with a human name and a spot on the company intranet or organizational chart may lead employees to wonder whether the company views software as a substitute for human judgment—or even as a potential replacement for their jobs. Employers can address these concerns by explaining what the agent is designed to do—and what it is not authorized to do—how it impacts specific tasks, and that human judgment and accountability remain essential. Involving employees in implementation, providing training, and communicating openly about these changes can help build trust and show how AI agents can support work rather than diminish employees’ contributions.