Prof. Aleksander Madry and graduate student Anish Athalye speak with BBC News reporter Linda Geddes about how AI systems can be tricked into seeing or hearing things that aren’t actually there. “People are looking at it as a potential security issue as these systems are increasingly being deployed in the real world,” Athalye explains.
MIT researchers have developed a way to prevent the theft of sensitive data hidden on a computer’s memory, writes TechCrunch’s Zack Whittaker. Storing sensitive data in different memory locations creates “clear boundaries for where sharing should and should not happen, so that programs with sensitive information can keep that data reasonably secure,” explains graduate student Vladimir Kiriansky.
Financial Times reporter Jemima Kelly reports that MIT researchers have developed a low-power chip that is hardwired to perform public-key encryption. Kelly writes that the chip is “potentially a game-changer for simple, low-powered products such as smart sensors used by industry to gauge things such as temperature and pressure, as well as health monitors.”
CSAIL Postdoc Gregory Falco writes in the San Francisco Chronicle about the importance of cybersecurity for satellites that interact with cars, televisions, and the internet. With new technological advancements like CubeSats, “we need to understand more about the legacy infrastructure on which these new technologies rely,” Falco concludes.
Engadget reporter Jon Fingas writes that MIT researchers have developed an encryption method that can secure sensitive data in neural networks without slowing machine learning systems. The method, notes Fingas, could “lead to more uses of internet-based neural networks for handling vital info, rather than forcing companies and institutions to either build expensive local equivalents or forget AI-based systems altogether.”
At the International Conference on Machine Learning, MIT researchers demonstrated an adversarial attack, fooling an AI system into thinking a 3-D printed turtle was a rifle, reports Matthew Hutson for Science. Prof. Aleksander Madry explains that the ability to fool AI systems shows, “we need to rethink all of our machine learning pipeline to make it more robust.”
In an article for The Wall Street Journal, Prof. Stuart Madnick writes about how companies can reduce their risk of cyberattacks by improving cybersecurity training and education among employees. “It’s crucial that support and enthusiasm for increasing cybersafety be visible at every level of the organization, from top executives and middle management to the individual,” explains Madnick.
Boston Globe Ideas columnist Kevin Lewis highlights a new study by MIT Sloan researchers that compares how well IT professionals and inexperienced students performed at a cybersecurity management game. Researchers concluded that because the experienced group didn’t perform better than the students, “training for better understanding of the complexities of cybersecurity is needed for decision-makers.”
In an article for Scientific American, Sloan research scientist Mohammad S. Jalali writes about the concerning ways hospitals are vulnerable to cyber attacks, despite being responsible for some of people’s most sensitive information. Jalali suggests that “policymakers, health care leaders and hospitals themselves should work together to make the industry as a whole less susceptible to attacks that threaten people’s privacy and their very lives.”
Forbes contributor Joe McKendrick describes how blockchain applications “may eventually reach into every corner of the business, providing online, ‘smart contracts.’” McKendrick cites research from the Media Lab’s Senior Advisor for Blockchain Opportunities, Michael Casey, whose recent book highlights the World Food Program’s use of blockchain for food distribution.
In a recent paper, CSAIL researchers propose a new system, Veil, which uses servers to conceal private web browsing data from internet service providers, reports Rob Verger of Popular Science. Through this system, the provider “would only see the connection to the blinding server, which hosts the actual content” and not the website the user is visiting, explains Verger.
Anant Agarwal, president of edX, and Kalyan Veeramachaneni, a principal research scientist at LIDS, are featured on Fast Company’s 2017 list of the “Most Creative People in Business.” Agarwal is celebrated for “mastering online education,” and Veeramachaneni for developing a system that enables humans and AI to work together to detect possible security threats.
Prof. Stuart Madnick writes for Forbes about the security threat posed by connecting household items to the internet. Madnick notes that it is, “great that computer-enabled internet-connect devices now bring wonderful new capabilities and conveniences. But there is also a need to take a broad view of the impact on our nation’s critical infrastructure.”
Joel Brenner, former NSA inspector general and a research fellow at MIT, speaks to BBC reporter Gareth Mitchell about an MIT report that examines cyber security threats to the nation’s infrastructure. “You can have a digital network that’s not public,” says Brenner, “but you shouldn’t be able to get to the controls of critical infrastructure through the public internet.”
A report from MIT’s Center for International Studies and CSAIL encourages the government to increase cybersecurity systems guarding the nation’s infrastructure, reports Jordan Graham for the Boston Herald. One suggestion from the report is to “establish incentives for owners and operators of private infrastructure who boost security,” explains Graham.