Credibility of Soft Law for Artificial Intelligence—Planning and Stakeholder Considerations

Seeing the CEOs
of four of the world’s largest tech companies appear before the U.S. Congress in July 2020 reinforced the idea that regulatory action for these tech giants is coming
[1]
. But a different subtext was also clear—specific regulation for future technologies that these and other companies have in research and development is likely a long way off, in no small part because these are fundamental platform technologies that easily cross geographic boundaries with uses, value, and risks that can be hard to predict. Work in Europe and China to regulate artificial intelligence (AI) may be further along, but will face similar challenges and limitations
[2]
,
[3]
. Companies, civil society, courts, and even governments will need to turn elsewhere for guidance and safeguards on the governance of AI, and society will need to turn elsewhere for effective safeguards. Most likely, these will come in the form of the array of multistakeholder initiatives, standards, codes of conduct, and the like that fall under the broad umbrella of “soft law”—mechanisms that “include various types of instruments that set forth substantive expectations but are not directly enforceable by government”
[4]
–

[6]
. The trust that society places in AI itself will depend on the credibility of the rules that guide its development and use, including those based on soft law. This is a tall order for rules built without state authority.