All Work Published on Privacy, Safety, Security
Safety Risks from Customizing Foundation Models via Fine-Tuning
This brief underscores the safety risks inherent in custom fine-tuning of large language models.
This brief underscores the safety risks inherent in custom fine-tuning of large language models.
A Stanford study reveals that leading AI companies are pulling user conversations for training, highlighting privacy risks and a need for clearer policies.
A Stanford study reveals that leading AI companies are pulling user conversations for training, highlighting privacy risks and a need for clearer policies.
Daniel E. Ho's Testimony Before the House Oversight and Accountability Cybersecurity, Information Technology, and Government Innovation Subcommittee
In this testimony presented to the U.S. House Subcommittee on Cybersecurity, Information Technology, and Government Innovation, Daniel E. Ho recommends six AI governance actions including boosting AI talent, establishing a mandated adverse event reporting system, and funding public AI research initiatives.
In this testimony presented to the U.S. House Subcommittee on Cybersecurity, Information Technology, and Government Innovation, Daniel E. Ho recommends six AI governance actions including boosting AI talent, establishing a mandated adverse event reporting system, and funding public AI research initiatives.
How Congress Could Stifle The Onslaught Of AI-Generated Child Sexual Abuse Material
HAI Policy Fellow Riana Pfefferkorn advises on ways in which the United States Congress could move the needle on model safety regarding AI-generated CSAM.
HAI Policy Fellow Riana Pfefferkorn advises on ways in which the United States Congress could move the needle on model safety regarding AI-generated CSAM.
The Privacy-Bias Trade-Off
This brief examines the “regulatory misalignment” of existing regulatory proposals for AI and urges policymakers to first consider feasibility, trade-offs, and unintended consequences before rushing into regulation.
This brief examines the “regulatory misalignment” of existing regulatory proposals for AI and urges policymakers to first consider feasibility, trade-offs, and unintended consequences before rushing into regulation.
HAI Policy Fellow Riana Pfefferkorn discusses scenarios when third parties might be able to access personal messaging data and how to keep those forms of digital communication private.
HAI Policy Fellow Riana Pfefferkorn discusses scenarios when third parties might be able to access personal messaging data and how to keep those forms of digital communication private.