Research Labs

Future of Identity Lab

From biometric privacy to ID cards for AI agents: who, or what, are we dealing with?

Introduction

Identity is being redefined on two fronts at once.

Human identity. Our identities increasingly live online, and some of the most personal data about us — our faces, voices and fingerprints — is being captured, analyzed and monetized, often without meaningful consent. The lab studies online identity and the protection of biometric data, and how individuals can retain control over what makes them identifiable.

The identity of AI agents. As AI agents increasingly act on behalf of people and companies, they urgently need an international registration system that gives each agent a unique, verifiable identity. Some countries are already moving in this direction: China, for example, has begun introducing a national standard for digital ID cards for AI agents.

With verifiable IDs, agents could be given more authority to conduct business autonomously, including signing contracts and making financial transactions. Technically, humans or companies own these agents. But it is conceivable that, as long as an agent delivers a healthy return, its owners will be content to let it operate with considerable freedom.

That raises the question of accountability. If an agent engages in illegal activity, who is held responsible? Traditional legal logic says the human or corporate owner remains responsible for what their AI does. Yet the way some AI developers publicly frame their latest models — as so capable that they are hard to control — challenges that notion, and risks shifting responsibility away from the companies that build and deploy them. If, in the near future, agents with official IDs come to be seen as separate legal entities, they could be held accountable directly. But the law would have to change: imprisonment is not an option, and a clone of an agent can be up and running in minutes. Do we take a "digital fingerprint" of an agent's code or harness, and sanction that?

Topics addressed in this lab

Recent and ongoing work

Protecting Persona Biometric Data: The Case of Facial Privacy

Hogenhout, L. and Wangmo, R.

Sept 2025

In short: Facial recognition data is being collected and monetized largely without meaningful consent, and today's privacy laws are too weak and fragmented to stop it. Comparing regulations across major jurisdictions, the paper argues for treating biometric data as an inalienable right rather than tradeable property.

Our labs