AI-enabled Manipulation and EU Law

45 Pages Posted: 28 Mar 2022

See all articles by Claire Boine

Claire Boine

University of Ottawa Faculty of Law; Artificial and Natural Intelligence Toulouse Institute

Date Written: June 1, 2021

Abstract

The European Commission, proposed three regulations containing provisions on digital manipulation. The first one, released in December 2020, is the Digital Services Act. It has a particular focus on the prevention of disinformation. The second proposed regulation that addresses digital manipulation is the Proposal for a Regulation of the European Parliament and of the Council on the transparency and targeting of political advertising, which was released in November 2021. The third proposed regulation we will address, the AI Act of April 2021, was set to adopt provisions addressing the risks of AI including the “specific risks of impersonation or deception.” A few weeks before the official release date, a draft of the proposal was leaked. It contained ambitious provisions against AI systems defined as manipulative. However, a few weeks later, when the actual proposed regulation came out, these provisions had been suppressed. The concept of manipulation had been replaced with the deployment of “subliminal techniques beyond a person’s consciousness in order to materially distort a person’s behaviour in a manner that causes or is likely to cause that person or another person physical or psychological harm.”

Although there is currently a surge in the collection of biometric and personal data to infer individuals’ personalities, preferences, and emotional states to influence their behaviors, the AI Act could be interpreted to be banning only a very narrow set of manipulative AI systems, and is likely to leave out some harmful manipulative systems. In the first part of this paper, we will discuss three issues that have been overlooked in the literature on manipulation using AI systems. In the second part, we will show the limitations of the EU Commission’s ban on subliminal techniques and expose its philosophical premises and contend that the Commission’s most recent guidelines on the UCPD provides hope for future directions. In the third part, we will propose a new definition of manipulation and legal solutions to the problem of AI manipulation.

Keywords: AI, manipulation, AI Act, UCPD, Digital Services Act, subliminal manipulation, Kant, Descartes, Cambridge Analytica, availability cascade, Babel technique, democracy, micro-targeting, political advertising

Suggested Citation

Boine, Claire, AI-enabled Manipulation and EU Law (June 1, 2021). Available at SSRN: https://ssrn.com/abstract=4042321 or http://dx.doi.org/10.2139/ssrn.4042321

Claire Boine (Contact Author)

University of Ottawa Faculty of Law ( email )

2292 Edwin Crescent
Ottawa, Ontario K2C 1H7
Canada

Artificial and Natural Intelligence Toulouse Institute ( email )

University of Toulouse
Toulouse
France

Do you have a job opening that you would like to promote on SSRN?

Paper statistics

Downloads
83
Abstract Views
182
rank
406,839
PlumX Metrics