Twitter granted researchers working for the Defense Department access to information shared by its users for study on combating online influence operations, a defense research program manager says.
Brian Kettler, a program manager at the Defense Advanced Research Projects Agency (DARPA), said unintentional collection of Americans’ data is possible.
However, there are processes in place to protect Americans’ personally identifiable information (PII) and prevent their data from being used by government-funded researchers, he added.
DARPA-funded researchers are following Twitter’s terms and conditions, Mr. Kettler said.
“We have an, the performer that’s in charge of data provisioning has an agreement with Twitter,” Mr. Kettler said. “Twitter has sort of, they scrutinized how we’re using the information, what information we’re accessing, so it’s all going through the platform T’s and C’s with a specific agreement.”
Two programs run by Mr. Kettler involve Twitter, namely the “Social Simulation for Evaluating Online Messaging Campaigns” (SocialSim) program and the “Influence Campaign Awareness and Sensemaking” (INCAS) program.
Mr. Kettler said that researchers on the SocialSim program have worked with Twitter and that those on the INCAS program intend to collect data from Twitter and plan to build algorithms to analyze data for foreign influence.
“You may inadvertently collect information on U.S. persons,” Mr. Kettler said. “When we do that we have procedures in place and policies to deal with that so that we are not intentionally collecting information on U.S. persons and if we unintentionally collect it that’s sequestered and not used. And we have a stringent policy in place in general to protect PII information that we’re collecting as well so we’re not looking at personal details of individuals.”
Mr. Kettler said the government is following Twitter’s rules to ensure it complies with the data it accesses on Twitter.
“For example, on Twitter, when somebody deletes a tweet, you have to delete the tweet in your database so it’s all going through the standard public ways you can procure data from these platforms,” he said. “We’re not scraping stuff, we’re working through the platforms.”
“Scraping” on social media websites refers to the use of automated software tools to access and extract or copy data from public profiles. The practice often runs afoul of tech platforms’ rules, but that does not stop hackers or foes from countries like China from scraping anyway.
Last year, American researcher Christopher Balding and Australian cybersecurity firm Internet 2.0 discovered that Chinese technology company Shenzhen Zhenua Data had surveilled the social media profiles of tens of thousands of Americans and collected information on people’s whereabouts and their personal and professional relationships.
Lawmakers have since debated several proposals on how to better protect the Americans’ data, including its export to foreign adversaries.
Twitter declined to comment for this report, including about what information the government-funded researchers are accessing.
Twitter’s privacy policy encourages users to think about what they decide to make public.
“You are responsible for your Tweets and other information you provide through our services, and you should think carefully about what you make public, especially if it is sensitive information,” reads Twitter’s website. “If you update your public information on Twitter, such as by deleting a Tweet or deactivating your account, we will reflect your updated content on Twitter.com, Twitter for iOS, and Twitter for Android. By publicly posting content, you are directing us to disclose that information as broadly as possible, including through our [application programming interfaces], and directing those accessing the information through our APIs to do the same.”
Asked whether Twitter users were made aware that DARPA researchers looked at their tweets, Mr. Kettler said Twitter treats DARPA researchers the same as anyone else it gives access.
Before the government’s selection of researchers for the INCAS program, DARPA required them to detail how they would control the collection of Americans’ data and apply “Human Subjects Research Controls” if they needed to do so, according to an October 2020 DARPA presentation.
The information accessed by the government-funded researchers looks to be tweets made available to the public.
A broad agency announcement about the INCAS program from October 2020 said “INCAS will primarily use publicly available data sources including multilingual, multi-platform social media, online news sources, and online reference data sources.”
“We hope that ultimately the tools that we develop to better understand the online information environment we can make open source and platforms can take advantage of but we’re not working with any platforms other than procuring their data through standard channels,” Mr. Kettler said.
