ProSocial Summit

A Virtual Event for Leaders Creating an Equitable and Safer Metaverse

Day 1

PST time zone
Wednesday, September 13th, 2023

8:10-8:30 am

Embracing the Future: Ensuring Real-Time Safety in the Digital Landscape

ActiveFence and Spectrum Labs have come together with the shared goal of enhancing the online experience for all by making the Internet a safer and more inclusive place. Ensuring online safety requires strategic preparation, advanced tools, comprehensive data, and the ability to identify malicious actors before they enter a platform and detect the ones that are already there. All of this helps trust & safety practitioners – the unsung heroes of the internet – construct digital environments that foster positive interactions.

With the rapid evolution of artificial intelligence and generative AI, the need for Trust & Safety at the LLM level need for Trust & Safety at the LLM level, cannot be overstated. This is where Nurdle plays a pivotal role by furnishing platforms with privacy-compliant data that serves as a foundation for the development and implementation of safer AI solutions.


Justin Davis
Noam Schwartz

8:30-9:00 am

Generative AI and the New Age of Trust & Safety



Dave Willner
Iftach Orr

9:00-10:00 am

Why 2023 is the Year That Humans and AI Change The (Online) World

While AI automation brings scalability and efficiency, it is crucial to acknowledge the inherent limitations of AI systems. The event will emphasize the importance of combining human moderation with AI automation to achieve optimal outcomes. Human moderators possess critical judgment, contextual understanding, and moral reasoning that AI models still lack. By integrating the strengths of both human and AI moderation, we can strike a delicate balance that upholds freedom of expression while curbing harmful content effectively.


Sharon Fisher
Head of Trust & Safety, Keywords Studios
Deniz Alkan
Senior Integrity Operations Manager, Bumble
Katie Zigelman
SVP of Sales & Solutions, Spectrum Labs
Barbara Konchinski
Director of Moderation, Minecraft
Nick Tapalansky
Senior Director of Safety & Support, MediaLab

10:00-10:30 am

The Trust & Safety Tech Stack - Best Practices for Product & Engineering Teams

Trust & Safety is a critical part of UX and product design, especially in platforms where users interact with each other and bad actors can bully, harass, and scam one another. Get a snapshot of how product managers, engineers, and data scientists can implement the right Trust & Safety solutions for their platform, which technologies are effective, and whether to buy it or build it themselves.


Carly Taylor



11:00-12:00 pm

The Future of Elections: The Role of Trust & Safety and the Impact of AI in Democratic Processes

ProTrust: Trust & Safety Best Practices
There are trust and safety challenges in any election process, and we are seeing those challenges increase with generative AI. We will aim to discuss existing disruptions to democratic processes observed on user-generated content platforms, potential harmful and beneficial uses of AI on elections, and to share some initiatives that are making headway in fostering civic integrity.


Patrica Cartes Andrés
Trust & Safety and Public Policy Advisor, Sotera
Lucia Gamboa
Senior Advisor, The Blue Owl Group
Edgard Oliveira
Emerging Threats & Risk Mitigation Expert at Kwai
Eric Davis
T&S, Security, Privacy Exec and advisor, previously at Google and Symantec
Grant Baker
Research Analyst for Technology and Democracy at Freedom House

11:00-11:30 am

Scaling Content Moderation

ProSolutions: Scaling TECH & DATA SOLUTIONS
Amidst the increasing complexities within the realm of content moderation, our aim is to discuss the strategies for content moderation orchestration. This exploration takes into account recent efforts to optimize resources and enhance efficiency. We also seek to delve into the potential of various technologies and methodologies to enhance the efficiency of content moderation.

We will also showcase our moderation control center. How to manage large moderation teams (local and remote), improve efficiency, productivity and utilization and ensure in real time that SLA for each policy area is met


Tomer Poran
VP Solution Strategy, ActiveFence
Assaf Ohana
CPO, ActiveFence
Matar Haller
VP of Data, ActiveFence

11:30-12:30 pm

How to Make LLMs Safer for Trust & Safety Through Personalization and Fine-Tuning

ProSolutions: Scaling TECH & DATA SOLUTIONS
In the rapidly evolving landscape of online interactions, ensuring Trust & Safety while enhancing user experience is paramount. Join this session to explore innovative approaches that combine the power of real, augmented, and synthetic data using Large Language Models (LLMs) for impactful Trust & Safety use cases.


Yuanbo Wang
machine learning engineer, NURDLE
Ryan Treichler
VP of product, spectrum labs
Jonathan Purnell
vp of data science, NURDLE
Patricia Thaine
Co-Founder and CEO, PRIVATE AI


The Future of Trust & Safety: Challenges and Best Practices

ProTrust: Trust & Safety Best Practices
This session explores the evolving landscape of trust and safety in digital platforms and provides insights into the key challenges and best practices involved in detection, moderation, prevention, and user education. We'll dive into the importance of alignment across teams and how collaboration enhances overall effectiveness of user safety. Additionally, we'll discuss measuring the success of Trust & Safety initiatives through analytics and reporting, highlighting the significance of data-driven decision-making. Lastly, we'll explore strategies for educating users, empowering them with the knowledge and tools to navigate digital platforms safely.


Matt Soeth
Head of trust & safety, Spectrum Labs
Alice Hunsberger
VP, Customer Experience, Grindr
Tami Bhaumik
VP of Civility and Partnerships, Roblox
Matt Murray
Director of Financial Crimes & Regulatory Affairs, GoFundMe
Jamie Sachs
Senior Manager, Vimeo

12:30-1:00 pm

Building a Robust Data Infrastructure for Detection Classifiers

ProSolutions: Scaling Tech & DATA SOLUTIONS
In today's data-driven world, the ability to effectively build and manage a data infrastructure is a critical skill for organizations aiming to stay competitive.

Our panel will share best practices for effectively managing, cleaning, and organizing data across your entire business and how to streamline data collection processes, optimize storage, and implement robust data strategies. Explore how partnering with a vendor can alleviate the burden by providing efficient data storage, processing capabilities, and data protection, ultimately leading to cost reductions and improved operational efficiency.


Othmane Rifki
Principal Applied Science, Spectrum Labs
Jonathan Purnell
VP of Data Science, NURDLE


Disruptions and Harms in Online Gaming

lunch keynote
The last 5-10 years have brought a wealth of investment in understanding and addressing disruption and harms in gaming and designing healthy, thriving spaces. While there is still much to be done as an industry and as a society, there are still a lot of learnings to be shared. This session will focus on a look at what we are seeing in games today, the research and best practices that have emerged, how we need to think about and design for digital thriving, and what’s next.


Kim Voll
Co-Founder, Fair Play Alliance + Head of Studio, Brace Yourself Games
Natasha Miller
Senior Research Scientist II, Blizzard
Eileen Shiue
Lead Metagames Systems Designer, Singularity 6

2:00-2:30 pm

Empowering Trust and Safety Teams with Threat Intelligence in Today's Changed Cyber World

ProTrust: Trust & Safety Best Practices
A discussion that will reshape the understanding of how threat intelligence and trust & safety weave together to redefine the digital landscape. Threat intelligence plays a crucial role in enhancing trust and safety operations by providing valuable insights into potential risks and adversaries. Threat intelligence can be a powerful tool for identifying and mitigating brand abuse-related risks.


Joe Azzouggagh
Trust & Safety manager, ruby
Tomer Poran
VP, Solution Strategy, activefence

2:00-2:30 pm

Scaling Data Preperation

ProSolutions: Scaling TECH & DATA SOLUTIONS
In this session, learn about the crucial aspects of data labeling and its impact on the quality of machine learning models. The speakers will explore advanced techniques in machine labeling and strategies to ensure high data label quality and quantity. They will also uncover innovative data generation and augmentation approaches specifically tailored for Training and Supervised Learning Trust & Safety applications.


Jonathan Purnell
Nina Lopatina
Christopher Guess

2:30-3:30 pm

Protecting Children Online: Approaches and Best Practices.

ProTrust: Trust & Safety Best Practices
In this session, learn about scaling online child safety content moderation and ensuring the safety of underage users in the digital landscape. The speakers will discuss key challenges faced by online platforms in addressing child safety concerns, navigating regional contexts and languages, and aligning cross-functional teams to comply with regulations such as COPPA, DSA, and Online Harms.

Our expert speakers will share their insights on detecting complex behaviors, including the identification and prevention of Child Sexual Abuse Material (CSAM), addressing bullying incidents, managing users under the age of 13 and 18, and educating underage users on proper online behaviors. We will explore effective strategies for engaging and retaining underage users through positive reinforcement and rewards for responsible digital conduct.


Stephen Balkam
Lisa Thee
managing director, data & ai, Launch COnsulting group
Dr. Avi Jager
head of child safety and human exploitation, activefence
Dr. Elizabeth Milovidov
Sr. Corporate Counsel,
The LEGO GrouP
Joel Silk
Senior Director of Trust & Safety Operations, Roblox

2:30-3:30 pm

Implications of LLMs on Trust & Safety Risks and Approaches for Safeguarding Them

ProSolutions: Scaling TECH & DATA SOLUTIONS
As Generative AI (Gen AI) rapidly advances and foundation models become prevalent across multiple industries, it is essential to monitor and reduce potential risks. Of major concern is that LLMs can be used to create dangerous content and provide advice for threat actors engaged in high-risk activities, from child exploitation to misinformation. During this session we will discuss the risks of LLMs and ways to keep them safe.


Inbal Goldberger
VP of trust & safety, activefence
Nitzan Tamari
Strategy Advisor, ActiveFence
Yoni Deblinger
VP of Extremism, ActiveFence
Guy Paltieli
Senior Child Safety Researcher, ActiveFence

3:30-4:00 pm

Digital Safety Regulatory Compliance

ProTrust: Trust & Safety Best Practices
New digital safety regulations around the world pose new requirements for online platforms. For example:
The DSA requires online platforms to establish ad-hoc and ongoing Trust & Safety mechanisms, to enhance their transparency and accountability.
This session will review some of the prominent regulations and will propose ways to comply using dedicated T&S platforms.


Michal Brand Gold
General Counsel, ActiveFence
Augustina Callegari
Project Lead, Digital
Safety, WEF
Assaf Ohana
CPO, ActiveFence
Laura De Boel
Partner, wILSON soNSINI


Closing remarks