AI Companies Propose 'Personhood Credentials' to Fight AI

AI companies like Microsoft and OpenAI are proposing personhood credentials as a way to fight online fraud, but the concept is drawing criticism for its potentially dystopian implications.

Share
AI Companies Propose 'Personhood Credentials' to Fight AI

Researchers from Microsoft and OpenAI, along with other contributors, have proposed the use of "personhood credentials" as a tool to counter fraud and misleading information online. According to Shrey Jain, a product manager at Microsoft, bad actors exploit anonymity to deceive others online, and AI is making it increasingly easy and cheap to create content indistinguishable from human-generated material.

Personhood Credentials as a Solution

As a solution to these challenges, Microsoft, OpenAI, and a number of academic researchers have proposed personhood credentials (PHCs). These credentials function as cryptographically authenticated identifiers given to individuals considered legitimate. Unlike traditional digital certificates, PHCs are meant to be pseudonymous in order to protect users' privacy.

A research paper, with more than 30 authors from institutions including Harvard, MIT, and Oxford, describes the concept. Personhood credentials are meant to protect users' privacy by being unlinkable and hard to trace across different services. Even so, the researchers acknowledge that PHCs can't protect against the extensive surveillance that already exists online.

Technical Implementation and Possible Issuers

PHCs are proposed as a framework rather than a specific technical solution, one where different organizations could offer these credentials. For example, US states could issue PHCs to people with a tax ID number, with or without a biometric link.

The researchers express concern that the internet isn't sufficiently prepared for the challenges AI could bring. Without proactive measures from the public, governments, and technologists, digital institutions risk being unprepared as AI-based agents are used by bad actors.

Criticism of the Proposal

Jacob Hoffman-Andrews, senior technologist at the Electronic Frontier Foundation, describes the proposal as "wildly dystopian." He warns that governments' historical inability to fairly issue such credentials could result in some people being denied their voice online. He also points to the risk that PHCs could become a tool for governments to control who is allowed to speak online, especially in countries where governments have an interest in large-scale disinformation.

Challenges to Overcome

The authors of the research paper acknowledge there are challenges with PHCs, such as ensuring fairness, supporting freedom of speech, and avoiding giving too much power to participants in the ecosystem. The systems also need to be robust against attacks and errors.

Despite the challenges that remain, PHCs aim to reduce large-scale fraud while protecting users' privacy and civil liberties.