We will study how to harden machine learning classifiers against adversarial attack. We will explore general mechanisms for making deep-learning classifiers more robust against attack, with a special focus on security for autonomous vehicles. Current schemes fail badly in the presence of an attacker who is trying to fool or…
The cybersecurity toolkitcollections of digital tools, tutorials, tips, best practices, and other recommendationshas emerged as a popular approach for preventing and addressing cybersecurity threats and attacks. Often these toolkits are oriented toward vulnerable populations who have unique and pressing needs related to cybersecurity but may not have access to the…
This researcher will undertake a residency and conduct an ethnography of Megvii, a startup company in Beijing that develops smart surveillance technologies, particularly facial recognition systems, using artificial intelligence techniques. Through interviews and ethnographic observations at this particular firm, the project will generate insights into how Chinese experts who currently…
Some relatively small states such as Estonia, Finland, Israel, and Singapore have become significant providers of national cyber-defense for their populations, regularly ranking alongside far larger states such as the US. This development is puzzling for security scholars as it represents a departure from the prevailing assumption that larger states…
Public high schools in the United States have increasingly digitized administrative and instructional systems over the last twenty years. Consequently, the security and welfare of students has been a primary concern, as reflected in regulations like the Federal Childrens Internet Protection Act (CIPA). Yet some cybersecurity measures intended for childrens…
The proliferation of automated decision-making systems has yielded much commercial success, but the potential of such systems to systematically generate biased decisions threatens to exacerbate the vulnerability of certain subgroups. Especially as the aim of machine learning algorithms shifts from making predictions for consumption by humans to making the very…
The growing use of cyber capabilities against civilian populations as a means and method of warfare is presenting new vulnerabilities and heightened risk profiles for refugees, internally-displaced populations (IDPs), and other civilian and civil society groups. However, humanitarian and human rights practitioners are hindered in their ability to keep pace…
Autonomous vehicles promise to save up to 30,000 lives per year and prevent 94% of all road accidents, according to the U.S. National Highway Traffic and Safety Administration. Yet, paradoxically, research suggests that regulations mandating algorithms that optimize for saving lives threaten to create a collective action problem that could…
As the proliferation of consumer Internet of Things (IoT) devices continues, so too do security problems that impact users, companies, and the Internet as a whole. But who is responsible when attacks from IoT-based threats like the Mirai Botnet cripple the Internet? This project examines the costs that insecure IoT…
The law governing the information economy is often not prescribed by legislators or courts, but rather by private entities using technology and standard-form contracts. This research proposal seeks to account for the often unobserved role of private ordering in the future of cybersecurity, and suggests that, similar to the revolutionary…