Roughly one year after Open AI's release of its publicly available generative artificial intelligence platform, ChatGPT, governments around the world struggle to implement firm guardrails to regulate it and other emerging artificial intelligence (AI) technologies.
The European Union, for example, had by December 8 reached a preliminary deal with its member nations that will likely require greater transparency from AI models based on their risk when the EU begins enforcing its "AI Act" around 2025. Meanwhile, in the United States, President Biden's October 30 executive order gave the U.S. Department of Labor and other agencies 180 days to develop AI-related guidance on monitoring employees, preventing worker displacement and encouraging transparency. Still, neither the European nor the U.S. advancement is expected to yield definitive rules anytime soon for employers to follow when using AI.
Current U.S. requirements for employers are few. As of this writing, only three jurisdictions—New York City, Illinois and Maryland—have enacted laws targeting employers' uses of AI in hiring by placing specific obligations on employers. New York City recently began requiring AI-bias audits. Illinois and Maryland, for their part, now require employers to obtain applicants' consent before using AI-driven video interviews. Employers considering applicants in those three jurisdictions should review their obligations before using any AI platform in hiring. Still, even those jurisdictions are not necessarily targeting the use of generative AI platforms like ChatGPT.
Without clear and applicable law, some employers may miss opportunities to see how AI products can help employees be more productive, while other employers might overlook the potential risks that come from allowing employees to use generative AI unfettered. But employers can both capture the benefits and mitigate their risks now by implementing clear employee-facing policies that allow for controlled use of AI while prohibiting riskier uses.
Employee handbooks and agreements may already prohibit employees from disclosing confidential customer or proprietary information into AI systems, including public generative ones like ChatGPT. Public generative systems may learn from and potentially regurgitate the information to other users – an important concern for employers with sensitive information.
But an AI-specific policy for employees can go further—designating AI systems or tools that the employer has vetted as being safe while simultaneously defining what types of AI platforms are prohibited under what circumstances. Prohibiting employees from using public and generative platforms for certain kinds of work would help prevent employee disclosure of trade-secret or other protected information. This way the company can benefit from the great advantages that AI tools can offer, while mitigating risk.
Also, regardless of whether they employ AI technologies, employers should ensure that their sensitive information is segmented behind protected databases with Terms of Use—ensuring that only employees with a business need to know can access it. In the future, many AI technologies may draw on established internal data sets available to employee-users of the technology. Without controlling who within an organization has access to what data, employers may inadvertently allow AI tools to disclose sensitive information to employees beyond those with a need to know.
Finally, employers must remember that existing employment laws such as Title 7 of the Civil Rights Act and the Age Discrimination in Employment Act already apply to their AI-based hiring and employment decisions. The Equal Employment Opportunity Commission (EEOC), for example, has begun enforcing these laws against employers that have used AI technology to sort resumes based on the protected characteristics of applicants.
No single answer applies to all employers. For some, clients or market forces may require using new technologies as they are released. Others still may prefer to avoid integrating the technology altogether. But without straightforward direction from their employers, employees may start using public generative AI for work and preemptively let the genie out of the bottle, causing their employers unnecessary exposure. Employers should strongly consider implementing an AI policy now that works for their business rather than waiting until the dust settles.
The content of this article is intended to provide a general guide to the subject matter. Specialist advice should be sought about your specific circumstances.
