Author: CARE Staff

On July 23, 2025, the Trump Administration published Winning the Race: AMERICA’S AI ACTION PLAN (“AI Action Plan”).[1]  Among the goals of the AI Action Plan is the elimination of inappropriate bias[2] and false information in the government’s AI systems.  The corruptions caused by inappropriate biasing, which may be introduced at any number of stages in what has been called “the AI pipeline,” lead to systems and outputs that are unreliable and, in some instances, injurious.[3]  A core concern regarding the outputs of such flawed systems is that the outputs can contain AI hallucinations – a phenomenon in which the system “in a large language model (LLM), often a generative AI chatbot or computer vision tool, perceives patterns or objects that are nonexistent or imperceptible to human observers, creating outputs that are nonsensical or altogether inaccurate.”[4] The roots of this concern are diverse and run deep, but a conspicuous and widely publicized example of hallucinating AI is provided by Google’s attempt to avoid bias and promote diversity in the outputs of its  generative AI Gemini model.  Google’s tampering with historical truths was ostensibly well-intentioned and harmless, but it pointed out some underlying AI-related concerns that have more serious and potentially injurious consequences.  These consequences arise when a flawed AI system’s output is being relied on to make decisions that affect people’s lives. The following text briefly recaps the Google Gemini matter and then discusses aspects of the AI Action Plan that are intended to identify and eliminate inappropriate bias and falsity in government AI systems.

*This article is the first in a series of discussions of Winning the Race: AMERICA’S AI ACTION PLAN, issued by the Trump Administration on July 23, 2025.

Attempts to exercise control over corporate artificial intelligence (“AI”) adoption, development, implementation, and management have surfaced in the form of Executive Orders, legislation, rulemaking, union and guild agreements, and case law. However, shareholders in publicly traded corporations have also used the shareholder Proposal and Proxy voting process to introduce sweeping or focused efforts to force corporate management to address and disclose information on a variety of AI-related issues. These Proposals are often couched in language that references ethical and significant social policy issues that arise in course of the target company’s AI-related business plans and activities. Of course, the determination of what is “ethical” or what constitutes a “significant social policy issue” can be quite subjective or driven by the advocate’s ancillary political and economic goals. This article discusses three recent attempts to use the shareholder Proposal and Proxy voting process to impose AI-related “transparency” and control requirements on major users of the technology. The common thread that runs through each effort is the strategic use of shareholder initiatives to attempt to influence the policies and management approach to AI and AI-related issues, including, in some instances, specific security, privacy, copyright, and personality rights issues.

Gary Rinkerman is a Founding Partner at the law firm of Pierson Ferdinand, LLP, an Honorary Professor of Intellectual Property Law at Queen Mary University School of Law in London, a member of George Mason University’s Center For Assurance Research and Engineering, and a Senior Fellow at George Mason University’s Center for Excellence in Government Cybersecurity Risk Management and Resilience. The views and information provided in this article express solely the work of author and do not comprise legal advice and are not for attribution to any entity represented by the author or with which he is affiliated or a member.

Gary Rinkerman is a partner at the law firm of FisherBroyles LLP, an Honorary Professor of U.S. Intellectual Property Law at Queen Mary University in London, and a Senior Fellow at the Center for Assurance Research and Engineering (“CARE”) in the College of Engineering and Computing at George Mason University, Virginia. For those interested in “digital archeology,” Professor Rinkerman also successfully argued one of the first cases in which copyright in object code was enforced and he co-founded and served as Editor-in-Chief for Computer Law Reporter, one of the first legal publications (in the 1980s) to focus exclusively on law and computer technologies. This article should not be considered legal advice. The presentation of facts and the opinions expressed in this article are attributable solely to the author and do not necessarily reflect the views of any persons, organizations or entities with which he is affiliated or whom he represents. The author would also like to thank J.P. Auffret, Director of CARE, for his continuing support and for his expertise in the frontier areas of Artificial Intelligence.