Document details

Red lines for AI: Resisting surveillance, reclaiming privacy

Article 19 (2024), 26 pp.
"This report identifies the contextual, social, and political factors that may shape decisions on whether to call for red lines on the use of artificial intelligence (AI) or whether to focus on other routes to challenging use cases, including through engagement with companies and the use of regulatory and judicial forums. The report finds that these are not necessarily either/or choices, in that campaigns to adopt red lines for biometrics are likely to benefit from parallel challenges to technological applications in other forums. Moreover, as social and political contexts shift, space may open for the pursuit of red lines on AI, including as a result of a moratorium on the use of particular technologies, successful outcomes in court, or the voluntary abandonment or temporary pause of the design, development, or sale of technologies by companies.
This report aims to equip civil society organisations with the information and considerations necessary to successfully pursue red lines in the specific contexts in which they work. It explores the current state of play, reflects on the current obstacles to having red lines adopted by governments, and identifies key factors for civil society organisations to consider at the time of determining their own strategies to establish red lines.
The report first identifies the three main reasons often given for banning specific technologies or use cases which underpin calls for red lines: inaccuracies in performance; inherent, unnecessary, or disproportionate risks to human rights which cannot be mitigated; and the exacerbation of power imbalances between institutions using face recognition and individuals. Second, the report maps where red lines on the use of AI have been secured, highlighting the variances in the scope and nature of such red lines in terms of technological coverage, actors, and geography. Third, the report offers recommendations for civil society and other actors to consider when deciding on whether and how to pursue red lines for biometrics, including through explicit advocacy campaigns, policy, and law reform or litigation." (Executive summary)
MAIN JUSTIFICATIONS FOR THE ESTABLISHMENT OF AI RED LINES, 5
Incompatibility with international human rights standards -- Inequality and power imbalances between institutions and individuals -- Inaccuracies in the technology
THE CURRENT STATE OF PLAY OF AI RED LINES, 8
AI red lines in legislation -- Measures which fall short of, but may lead to, the adoption of AI red lines
OUR RECOMMENDATIONS FOR STRATEGY DEVELOPMENT AND POSITIONING ON AI RED LINES, 13
1 Incorporate demands for state accountability and transparency as part of AI red lines strategies, 14
2 Develop a clear and multidimensional narrative on the need for AI red lines, 16
3 Carefully assess the potential for pitfalls or blowback when identifying advocacy opportunities, 18
4 Determine the type and scope of the AI red lines, 19
5 Determine the geographic scope of the AI red lines, 21
CONCLUSIONS, 22