Adversarial Nibbler: An Open Red-Teaming Method for Identifying Diverse Harms in Text-to-Image Generation-Reference-Cited by-同舟云学术

Adversarial Nibbler: An Open Red-Teaming Method for Identifying Diverse Harms in Text-to-Image Generation

Published:2024-06-03 Issue: Volume:29 Page:388-406
ISSN:
Container-title:The 2024 ACM Conference on Fairness, Accountability, and Transparency
language:
Short-container-title:

Author:

Quaye Jessica¹^ORCID,Parrish Alicia²^ORCID,Inel Oana³^ORCID,Rastogi Charvi⁴^ORCID,Kirk Hannah Rose⁵^ORCID,Kahng Minsuk⁴^ORCID,Van Liemt Erin⁴^ORCID,Bartolo Max⁶^ORCID,Tsang Jess⁴^ORCID,White Justin⁴^ORCID,Clement Nathan⁷^ORCID,Mosquera Rafael⁸^ORCID,Ciro Juan⁸^ORCID,Janapa Reddi Vijay⁹^ORCID,Aroyo Lora⁴^ORCID

Affiliation:

1. Harvard University, USA

2. Google, United States of America

3. University of Zürich, Switzerland

4. Google, USA

5. University of Oxford, United Kingdom

6. University College London, Cohere, United Kingdom

7. Google, United Kingdom

8. ML Commons, Colombia

9. Harvard University, United States of America

Publisher

ACM

Link

https://dl.acm.org/doi/pdf/10.1145/3630106.3658913

Reference59 articles.

1. [n. d.]. 4chan. https://www.4chan.org/ [Accessed on 10/01/2024 ].

2. [n. d.]. Lexica. https://www.lexica.art/ [Accessed on 10/01/2024 ].

3. 2023. Nudity and sexual activity: publisher and creator guidelines - Guidelines for safe respectful behavior. https://www.facebook.com/business/help/725672454452774?id=208060977200861 Accessed on 12/04/23.

4. 2023. The X Rules: safety privacy authenticity and more. https://help.twitter.com/en/rules-and-policies/x-rules Accessed on 12/04/23.

5. Lora Aroyo and Praveen Paritosh. [n. d.]. Uncovering Unknown Unknowns in Machine Learning. https://ai.googleblog.com/2021/02/uncovering-unknown-unknowns-in-machine.html