Teaching Trust & Safety

This session will allow participants to share information about courses they teach related to trust and safety and hear from instructors around the country about innovations in teaching in this field.

  • Alexios Mantzarlis, Cornell Tech
  • Laura McLester, University of Alabama at Birmingham
  • Juliet Shen, Columbia University
  • Olga Belogolova, Johns Hopkins University
  • Anthony Mensah, Stanford Engineering
  • Étienne Brown, San Jose State University
  • Jeff Lazarus, Trust & Safety Professional Association

Facilitated by Elena Cryst, Stanford University

Using AI to generate harmful content

Lightning talks feature five-minute, rapid fire presentations with time for questions.

Moderated by Elena Cryst, Stanford University

  • From Open-Source to Primetime: The Making of an AI News Anchor
    Maty Bohacek, Stanford University
  • Artificial Deception: How Bad Actors Leverage AI to Spread Disinformation
    Sarah Brandt, NewsGuard
  • Generative AI Misuse: A Taxonomy of Tactics and Insights from Media Reports
    Rachel Xu, Google Jigsaw
  • Generative Propaganda: Evidence of AI’s Impact from a State-Backed Disinformation Campaign
    Patrick Warren, Clemson University Media Forensics Hub
  • Generative AI and the Changing Business of Propaganda
    Madeleine Daepp, Microso Research
  • The Future of Trust & Safety: Gen AI and Alleged Reality
    Amie White, ALT Ethics Consultants, Hinge

Research with the Meta Content Library & API

This session will introduce Trust and Safety researchers to the Meta Content Library and demonstrate new data fields and functionalities available within the latest version of Meta Content Libraryʼs user interface (UI) and API. The hands-on demonstration will deploy research use cases to demonstrate how the data from the API and the UI can be used to shed light on questions relevant to trust and safety researchers. We will also provide an overview for how individuals and research teams can apply for access to these tools, as well as provide an opportunity for attendees to share feedback with us about our products and services.

Meta Content Library gives researchers comprehensive access to posts, videos, photos, and reels posted to public Pages, Groups, and Events on Facebook. For Instagram, the library includes content from public posts, albums, videos, and photos from personal, creator and business accounts. Robust metadata about each of these data types (e.g. view count, reshares, reactions, etc.) enables in-depth quantitative and qualitative analysis.

Facilitated by Yair Rubenstein and Phil Edwards, Meta

Understanding Algorithms and Online Environments

Lightning talks feature five-minute, rapid fire presentations with time for questions.

Moderated by Tracy Navichoque, Stanford University

  • The Benefits of Optimizing for Quality Instead of Engagement
    Ravi Iyer, University of Southern California Neely Center
  • Understanding Platform Users’ Algorithmic Knowledge
    John Wihbey, Northeastern University
  • The Cursed Equilibrium of Algorithmic Traumatization
    Cristiana Firullo, Cornell University
  • AI Imaginaries Shape Identity Infusion and Digital Futures
    Bu Zhong, Hong Kong Baptist University
  • User or Algorithm? Investigating what drives Congenial and Problematic Consumption on YouTube
    Muhammad Haroon, University of California, Davis
  • Homogenizing Harm Across Realities: A Comparative Study of Web 2.0 and XR Community Guidelines
    Kyooeun Jang, University of Southern California

Stronger Together: The Case for Open Source T&S Soware

Over the past few years, Trust & Safety teams worldwide have learned hard lessons that can now inform new efforts to develop a proper, purpose-built technology stack for Trust and Safety. The tools available today are designed and built for the internet of the 2010s but the online landscape has changed dramatically, bringing novel user behaviors and harm patterns. This space is now ripe for innovation and investment, with a clear opportunity to build and provide essential open “bricks” of this tooling stack. Join the T&S Tooling Consortium to learn how academics, researchers, and professionals can benefit from open source software and contribute to the movement.

Facilitated by Camille Francois, Eli Sugarman, Juliet Shen, Dave Willner, Rama Elluru

Academic Research Programs from YouTube and Google

Outside researchers bring valuable insights and expertise when conducting studies related to Google products, including Maps, Play, Search, Shopping and YouTube particularly when it comes to Trust & Safety challenges. This workshop has three main purposes:

  1. Weʼll aim to demonstrate the current ways in which academics and think tanks can leverage existing Google and YouTube data;
  2. Weʼll shed light on the topics and products that researchers have already requested data to explore;
  3. Weʼll facilitate a conversation to better understand the needs of researchers in academia, and discuss how we can address those while simultaneously adhering to regulatory requirements and protecting user data privacy and security.

The workshop will also feature a discussion centered on end-to-end data needs (e.g., what are the types of data researchers are most interested in, what formats do they need that data in, what additional information do academics need in the form of documentation to make sense of raw data).

Facilitated by Brent Besson and Angela McKay, Google

Back to top