On AI & personhood credentials: a new paper
paper title
Personhood credentials: Artificial intelligence and the need for privacy-preserving ways to distinguish who is real online
authors
Steven Adler, Zoë Hitzig, Shrey Jain, Catherine Brewer, Wayne Chang, Renée DiResta, Eddy Lazzarin (a16z crypto), Sean McGregor, Wendy Seltzer, Divya Siddarth, Nouran Soliman, Tobin South, Connor Spelliscy, Manu Sporny, Varya Srivastava, John Bailey, Brian Christian, Andrew Critch, Ronnie Falcon, Heather Flanagan, Kim Hamilton Duffy, Eric Ho, Claire Leibowicz, Srikanth Nadhamuni, Alan Z. Rozenshtein, David Schnurr, Evan Shapiro, Lacey Strahm, Andrew Trask, Zoe Weinberg, Cedric Whitney, Tom Zick
author affiliations include: OpenAI, Harvard Society of Fellows, Microsoft, University of Oxford, SpruceID, a16z crypto, UL Research Institutes, Tucows, Collective Intelligence Project, MIT, Decentralization Research Center, Digital Bazaar, American Enterprise Institute, Center for Human-Compatible AI, UC Berkeley, OpenMined, Decentralized Identity Foundation, Goodfire, Partnership on AI, Governments Foundation, University of Minnesota Law School, Mina Foundation, Ex/Ante, Berkman Klein Center for Internet & Society
paper abstract
Anonymity is an important principle online. Yet, malicious actors have long used misleading identities to conduct fraud, launch cyberattacks, spread disinformation, and carry out other deceptive schemes. With the advent of increasingly capable AI, bad actors now have the potential to amplify the scale and effectiveness of their operations, intensifying the challenge of balancing anonymity and accountability online.
Existing countermeasures to automated deception – such as CAPTCHAs and anomaly detection – are inadequate against sophisticated AI, while stringent identity verification solutions require personal information, undermining anonymity. This paper underscores the critical need to empower users and online services with privacy-preserving strategies that counter large-scale online deception. This need stems from two trends: AI’s increasing indistinguishability (i.e., lifelike content and avatars, agentic activity) from people online, and AI’s increasing scalability (i.e., cost-effectiveness, accessibility).
We focus on a particular countermeasure: “personhood credentials,” digital credentials that enable users to verify that they are real people – not AIs – to online services, without disclosing any personal information. Related proposals have a long history, from anonymous credentials to other limited proof-of-personhood schemes, which provide both technical foundations and practical insights that a personhood credential system could draw upon. A personhood credentialing system, according to our definition, does not need to be global or biometrics-based. These tools could, without compromising privacy, preserve the basic usability of the Internet in the face of highly capable AI – affording more trustworthy online spaces for people and helping reduce recurrent policy abuses by bad actors. After surveying the benefits, we also examine deployment risks and design challenges. We conclude with actionable next steps that incorporate public feedback for policymakers, technologists, and standards bodies to consider. [link to full paper]
our summary
The paper underscores the critical need to empower users and online services with privacy-preserving strategies that counter large-scale online deception. The paper addresses the issue of how AI is leading to a proliferation of misleading content – for example: scams; impersonating others/ agentic activity; Sybil attacks/ multiple pseudonymous identities generated by a single attacker; realistic/ lifelike AI-generated content intended to deceive; and so on.
The paper focuses on how “private personhood credentials” (PHCs) are one of the most promising countermeasures to deal with this issue. Such digital credentials enable users to verify that they are real people – not AIs – to online services. And importantly: without disclosing any personal information.
The broader category of work here is also related to the important concept of “proof of personhood” — that is, a mechanism that digitally verifies an individual’s humanness and uniqueness — however, that would require a more global, biometrics-based approach to a personhood credentialing system.