Self-Destructing Models: Increasing the Costs of Harmful Dual Uses of Foundation Models

Peter Henderson, Eric Mitchell, Christopher Manning, Dan Jurafsky, Chelsea Finn

Research output: Chapter in Book/Report/Conference proceedingConference contribution

1 Scopus citations

Abstract

A growing ecosystem of large, open-source foundation models has reduced the labeled data and technical expertise necessary to apply machine learning to many new problems. Yet foundation models pose a clear dual-use risk, indiscriminately reducing the costs of building both harmful and beneficial machine learning systems. Policy tools such as restricted model access and export controls are the primary methods currently used to mitigate such dual-use risks. In this work, we review potential safe-release strategies and argue that both policymakers and AI researchers would benefit from fundamentally new technologies enabling more precise control over the downstream usage of open-source foundation models. We propose one such approach: the task blocking paradigm, in which foundation models are trained with an additional mechanism to impede adaptation to harmful tasks without sacrificing performance on desirable tasks. We call the resulting models self-destructing models, inspired by mechanisms that prevent adversaries from using tools for harmful purposes. We present an algorithm for training self-destructing models leveraging techniques from meta-learning and adversarial learning, which we call meta-learned adversarial censoring (MLAC). In a small-scale experiment, we show MLAC can largely prevent a BERT-style model from being re-purposed to perform gender identification without harming the model's ability to perform profession classification.

Original languageEnglish (US)
Title of host publicationAIES 2023 - Proceedings of the 2023 AAAI/ACM Conference on AI, Ethics, and Society
PublisherAssociation for Computing Machinery, Inc
Pages287-296
Number of pages10
ISBN (Electronic)9798400702310
DOIs
StatePublished - Aug 8 2023
Externally publishedYes
Event2023 AAAI / ACM Conference on Artificial Intelligence, Ethics, and Society, AIES 2023 - Montreal, Canada
Duration: Aug 8 2023Aug 10 2023

Publication series

NameAIES 2023 - Proceedings of the 2023 AAAI/ACM Conference on AI, Ethics, and Society

Conference

Conference2023 AAAI / ACM Conference on Artificial Intelligence, Ethics, and Society, AIES 2023
Country/TerritoryCanada
CityMontreal
Period8/8/238/10/23

All Science Journal Classification (ASJC) codes

  • Artificial Intelligence

Fingerprint

Dive into the research topics of 'Self-Destructing Models: Increasing the Costs of Harmful Dual Uses of Foundation Models'. Together they form a unique fingerprint.

Cite this