Skip to main content

UMD Cybersecurity Forum: AI’s Role in Enforcing Online Safety Policy

Back to All News
IA Conference

Common business priorities of brand protection and mitigating liability are especially challenging when you’re Facebook hosting two billion-plus users around the globe, who generate billions of posts a week in more than a hundred languages.

Facebook’s Head of Global Safety Antigone Davis outlined her company’s approach to protecting its users, as the luncheon speaker for the Annual Forum on Financial Information Systems and Cybersecurity: A Public Policy Perspective, cohosted on Jan. 9, 2019 by the Robert H. Smith School of Business and School of Public Policy at the University of Maryland.

In introducing Davis, UMD Public Policy Dean Robert Orr noted Davis’ previous role as senior advisor on online safety and privacy to the Maryland Attorney General “in building the policy infrastructure necessary to grapple with these issues at the state level.” With Facebook, she works with a policy team of legal and PR experts whose recently published Community Standards targets users and guides the company’s 7,000-plus moderators who decide whether to remove content reported by users or generated via artificial intelligence (AI).

One of the forum’s participants asked Davis to describe, from a Facebook standpoint, the role of AI in the online safety/privacy-policy engine going forward. AI is especially effective against child exploitation, said Davis, referring to the recently released Facebook Transparency Report. Facebook is using “behavioral classifier” algorithms, which determine the nature of images and conversations, to identify “inappropriate interactions with children.” Subsequently, in the second quarter of 2018, Facebook removed 99 percent of 8.7 million pieces of child exploitation content before it was user-reported.

Facebook’s AI-initiated removal rate of hate speech content, on the other hand, has been low but climbing -- from about 20 to 60 percent through the first three iterations of the transparency enforcement report. “Technology will get better and better, as you can feed it more data. But, there will always be that place where we need human [interpretation],” Davis said.

This was the 15th staging of the annual conference organized by Lawrence Gordon, EY Alumni Professor of Managerial Accounting and Information Assurance, in the Smith School; Martin Loeb, professor of accounting and information assurance and Deloitte & Touche LLP Faculty Fellow, in the Smith School; and William Lucyshyn, research professor in the School of Public Policy and Director of Research for the school's Center for Public Policy and Private Enterprise.  

Gordon’s and Loeb’s seminal work in cyber risk management was alluded to in Maryland Smith Dean’s Chair of Management Michael O. Ball’s conference-opening remarks. “[Maryland Smith] has had a very strong research culture…We’ve been ahead of the curve when you consider, for example, the  Gordon-Loeb Model as a cybersecurity investment framework and the Center for Health Information and Decision Systems and its focus on digital technologies to improve healthcare.”

Other conference presenters included:

Shaun Wang, professor and director of the Insurance Risk and Finance Research Center at Nanyang Technological University, who overviewed Singapore's Cyber Risk Management Project — a university-government-industry partnership that he directs with the Monetary Authority of Singapore and the insurance industry.

Stacey Ferris of RM Advisory Services discussed his Economic Impact Study of the Advanced Encryption Standard. Advanced Encryption Standards, or AES, is one of the most widely used methods for encrypting and decrypting sensitive information.

Shouhuai Xu, professor and Director of Laboratory of Cybersecurity Dynamics at the University of Texas-San Antonio, presented his “Cybersecurity Dynamics ” paper.

UMD School of Public Policy Associate Research Professor Charles Harry described the Maryland Global Initiative for Cybersecurity (MaGIC), under his direction, and its campuswide mission to promote cyber education, research and development activities.

Gerasimos J. Stellatos, a principal in PwC’s Cyber Incident Threat Management practice, addressed the role of cybersecurity in accounting, in delivering the Ira H. Shapiro Memorial Lecture.

Notable Software CEO Rebecca Mercuri gave insight to government prosecutions of cybercrime, drawn from her work and expert testimonies as a widely respected computer forensics investigator.  


For Media Inquiries:
Megan Campbell
Senior Director of Strategic Communications
For More from the School of Public Policy:
Sign up for SPP News