A dataset developed by a US government research body to benchmark facial recognition systems contained photographs of journalists, writers and political activists without their consent and potentially used for military applications.

According (pdf) to US government agency Iarpa, 'the Janus program dramatically improved the performance of facial recognition software by increasing the speed and accuracy of identity matching.' 

But the dataset included a number of political activists, civil rights advocates, and journalists, including Ai Wei Wei, Tracey Emin, Evgeny Morozov, John Maeda, and Ta-Nehisi Coates, according to activist Adam Harvey and highlighted by a 2019 Financial Times investigation.

Mone of these individuals had been made aware of their inclusion in the database by Noblis or Iarpa, and their images had been obtained without their explicit consent. Furthermore, the use of YouTube videos constituted a clear violation of the platform's terms of service. 

In addition, as the FT pointed out, the use of the dataset by companies such as Chinese AI firm SenseTime and Japanese IT firm NEC, and by organisations such as China's National University of Defense Technology, raised concerns about its potential use for military and security purposes, including the mass surveillance of Uyghurs and other oppressed minorities.

The findings prompted ethical concerns about the need for greater oversight and ethical considerations in the development and use of facial recognition datasets and technologies.