IEEE Technology and Society Magazine - December 2021 - 96

Soft Law Governance of Artificial Intelligence
those constraints by outsourcing to companies.
Government abuses of commercially available technologies
would undermine the credibility of any
soft law governance of that sector. Thus, credible
soft law frameworks for FRT will require hard law
restrictions on how government can obtain and use
commercial FRT.
Hard law restrictions on law enforcement use
would also provide clearer guidance to commercial
entities deploying FRT for purposes of safety
and security. For instance, what if an FRT system
flags a customer as a known shop lifter and prompts
the store manager on site to call the police? Clarity
about how police can or cannot use that information
in prosecution of the customer will help companies
make better-calibrated choices about when to
involve law enforcement.
These concerns apply to governments at all levels,
including internationally. Many governments
lack the U.S.'s constitutional restrictions on data use.
Some companies have voluntarily limited shipments
of tech to oppressive regimes. Hard law restrictions
on the sale of FRT to certain foreign governments
may be appropriate. Yet, individuals under oppressive
regimes may find liberty-enhancing uses for FRT
(recall U.S. and Hong Kong protesters using FRT
to identify abusive police officers), suggesting that
blanket bans on exporting the technology could
have costs, as well [21].
Examples of hard law approaches
State and local legislatures provide some templates
for how hard law approaches to law enforcement
could protect civil liberties and help soft law
commercial FRT approaches maintain credibility.
Some cities have issued flat bans on law enforcement
uses of FRT. According to the Lawfare blog,
" Eight cities in California and Massachusetts have
banned government use of facial recognition altogether,
while Portland, Oregon, as mentioned earlier,
has banned both public- and private-sector use
of the technology. Three states have banned the
deployment of facial recognition in police body
cameras " [69]. This list seems likely to grow. Flat
bans, however, are a brittle governance approach.
They prohibit even compelling use-cases. If a tragically
missed opportunity-such as a terrorist
attack-triggers the withdrawal of such a ban, the
remaining governance structure will be underdeveloped
and untested.
96
The Facial Recognition Technology Warrant Act,
introduced by Senators Coons (D-Del.) and Lee (R-UT)
provides a better example of a hard law approach to
government use [70]. That bill would require federal
law enforcement to get a probable cause warrant to
use FRT to track an individual for more than 72 h. The
warrant requirement does not apply to a single identification
of an individual, if there is no subsequent
attempt to track the individual. This type of approach
allows the deployment of the technology in the law
enforcement toolbox, but under constraints that are
consistent with past practice.
commercial uses of FRT would benefit from a soft
law governance approach, but government uses need
hard law constraints. Unfortunately, many critics of
FRT and related technologies focus on regulation of
commercial uses and fail to distinguish commercial
uses from government uses. This has occasionally
led to legislation, such as Illinois' BIPA, that applies
inflexible hard law restrictions on commercial uses
while entirely exempting government applications.
This is perversely driving FRT business models and
investment to overemphasize government and law
enforcement use cases, where the risks of harm from
flawed technology are much higher. To ensure that
the beneficial commercial uses of FRT can continue
to develop under a soft law framework, we need to
protect civil liberties-by using hard law to constrain
government FRT uses.

Acknowledgment
This work was supported by the authors' employer
as part of their normal duties. We are extremely
grateful for Parker Kobayashi's research assistance
on this article.
 References
[1] S. D. Warren and L. D. Brandeis, " The right to
privacy, " Harvard Law Rev., vol. 4, no. 5, pp. 193-220,
Dec. 1890.
[2] I. R. Kramer, " The birth of privacy law: A century
since warren and brandeis, " Catholic Univ. Law Rev.,
vol. 39, no. 3, p. 703, 1990. [Online]. Available: https://
scholarship.law.edu/lawreview/
[3] Carpenter v. United States, 2018, p. 2206, vol. 138.
[4] Maryland Wiretapping and Electronic Surveillance
Act, Maryland Gen. Assembly, May 1977.
[5] G. Marchant, " Soft law governance of artificial
intelligence, " AI Pulse, Jan. 25, 2019. [Online].
IEEE Technology and Society Magazine
https://scholarship.law.edu/lawreview/ https://scholarship.law.edu/lawreview/
IEEE Technology and Society Magazine - December 2021

Table of Contents for the Digital Edition of IEEE Technology and Society Magazine - December 2021
