Artificial intelligence (AI) and machine learning applications like ChatGPT are the hot new technologies expected to changes our lives fundamentally. Taking the human element out of monotonous tasks promises to increase efficiency. But history shows that technological developments always result in unanticipated circumstances and unexpected results. What are the potential legal implications of AI and machine learning? Even at this early stage of development, these technologies preseent serious legal issues.

Media and Content Creators

Twitter and the internet are rife with stories of aspiring content creators using ChatGPT and other AI technologies to create content. Where a medieval monk might have required six to twelve months to transcribe a book, a teenager can use ChatGPT to create a book in a few hours without leaving their bedroom. When almost anyone can create great quantities of digital content, there will necessarily be massive legal ramifications in the world of intellectual property. The U.S. Copyright Office has already take the position that copyright protection is not available for works created by a computer algorithm.^[1] But this ruling leaves open huge gaps in the law. Even if an AI-created work is not protectable by copyright laws, it still exists. Without copyright protection, presumably that work is now part of the public domain and anyone can use and copy it freely. AI related technologies may give rise to a massive increase of content that is not protectable under existing law and free for use by anyone. But freedom to use and copy does not mean freedom from liability. If by chance some portion of an AI-created content directly copies or is derivative of a protected work, then the publisher of the AI-created work may be liable. Even innocent copying may result in liability under existing law.

Employment Law

It is unlawful for employers to discriminate against job applicants based on race, religion, national origin, color, pregnancy, or sex (including gender). There is an entire body of case law addressing whether certain conduct by an employer constitutes unlawful discrimination against potential employees based on one of these criteria. Under this case law, discrimination may be shown by proving disparate impact on a protected class. Disparate impact liability does not require intentional discrimination.

Employers seeking to avoid a claim of discrimination by relying on AI technology to screen job applicants should think twice. It is not clear that current AI technology is ready for the job. With the public launch of ChatGPT and other mass-market AI software, numerous journalists and others have reported evidence of inherent bias in the current technology. An employer who implements this technology could face a claim of discrimination if the results of algorithmic screening of job candidates results in disparate impacts on protected classes. Evidence of the employer’s ignorance of the flaws in the system is not likely to be a valid defense. Even if an employer acted entirely in good faith in implementing an AI based candidate screening system, that employer can still be liable for discrimination if a plaintiff demonstrates that the AI screening system resulted in a disparate impact.

Civil Rights

AI technology is already haveing an impact on civil rights. Facial recognition technology, for instance, promises a quick and efficient tool to identifyand locate criminal suspects. But there are substantial concerns about its accuracy. For example, in January 2020, the Detroit Police Department mistakenly arrested Robert Williams for theft based on a “match” by the department’s facial recognition technology. Williams, represented by the ACLU, is now suing the Detroit Police Department for wrongful arrest and violations of the Fourth Amendment and Mr. Williams’ civil rights.^[2]

Someday, facial recognition technology may reach the same level of accuracy as fingerprinting and DNA, but that day is not today. The current limitations of AI and machine learning technology suggest that it is too soon to rely on these technologies in the criminal justice system, where we as a society have long held the value that it is “better that 10 guilty persons escape, than that one innocent suffer.”^[3] Law enforcement agencies who rely too much on bleeding edge technologies will likely face liability under the civil rights laws.

The potential benefits of AI and machine learning technologies are promising. The unexpected outcomes and risks will likely keep courts and policymakers busy for years to come.