320 R. Vecellio Segate focusing on credentials for user authentication, with an emphasis on security, interoperability, and robustness.6 Many other variables, however, seem to be left to the periphery, including matters of cultural sophistication, data transfer, and free informed consent. Interfaces with artificial intelligence (AI) are also of concern, particularly if machine learning is deployed to validate profiles’ “alignment” with preset expectations and/or learn from a user’s patterns of engagement with the identity provider. In the hands of ill-intentioned agencies, algorithms might be trained to systematically discriminate fractions of the population, or to flag up individuals’ profile for further scrutiny by hostile authorities – in this event, individuals would not be prevented from onboarding; to the contrary, they would be even prioritised in order to be more rapidly “handled” (i.e. persecuted) by enemy factions, semi-privatised militias, or political opponents. Data patterns may end up feeding malevolent “risk profiling” that pursues identity onboarding for the only (or main) sake of censing (and censoring) regime challengers. Of course, not all such intents are malevolent: some may well adhere to genuine public-policy purposes, such as scaling-down the underground economy, preventing violent crime, or enhancing the safety of neglected neighborhoods; it all depends on the transparency and accountability record of those implementing AI solutions for biometrics. And when it comes to regulating AI applications, just like with any technology that impacts and records human movement, techno-specific regulation is not a panacea in isolation; ‘parallel efforts [are warranted] to set up a global institutional framework adequate to address the present and future global governance issues’ (Neuwirth 2024). 3 Mediating Trust Between the Centre and the Peripheries: Shall we Trust the Introducer? Once a role devised by banks only, the introducer has recently captured the attention of public lawyers, policymakers, and sociologists for its application in the domain of technology-aided identity onboarding. For the sake of our discussion here, the introducer is defined as a mediating party that is trusted by both state authorities and identity-stripped individuals in order for the latter to prove who they are or to be provided with the required technical assistance to do so themselves. This second scenario (the introducer as a technology enabler and capacity builder) raises no substantial controversy: choosing introducers would still be challenging, but their 6 In plain language, “robustness” is the multidimensional ability of a computing system to deal with errors (failures, noise, erroneous training) in such a way as to preserve the integrity and validity of the output data (Ding, Janssen, and Crowcroft 2021, 165–166; Maple et al. 2021a, 10–12; 21).