Twitter gave back-end approval to government-run accounts working to influence public opinion on matters of foreign policy at the behest of the Pentagon. The revelation comes in the latest batch of “Twitter Files,” granted to The Intercept’s Lee Fang.
According to the report, as far back as 2017, U.S. Central Command (CENTCOM) requested that Twitter “whitelist” a series of Arab-language accounts the department said they used “to amplify certain messages.” At the time, Twitter had fortified content filters aimed at identifying and removing content associated with terrorist groups based in the Middle East, including ISIS. Military-run accounts that frequently engaged with the content were being caught in their filtration and visibility reduction systems.
Several of the accounts’ content focused on promoting and framing U.S. military activity in the Middle East. While some accounts disclosed their affiliation with the U.S. government, The Intercept found that others deleted their disclosures at some point. According to former employees speaking under condition of anonymity, the “whitelist” internal tag applied to these accounts essentially gave them all of the benefits of verification, without a visible verification tag on the account.
At least one account identified by The Intercept as whitelisted in CENTCOM’s initial 2017 request was identified in a Stanford Internet Observatory report as an account using a deep fake as a profile image, and that was part of a vast state-backed online influence operation. The account, “@mktashif,” purported to be an unbiased news account “dedicated to serving Iraqis and Arabs.” The account was created using a CENTCOM-affiliated email.
Twitter’s cooperation with national security agencies running propaganda operations on the platform, contradicts Twitter’s long-held stance on foreign influence operations. In a 2020 appearance before Congress, Twitter spokesperson Nick Pickles stated that the platform’s goal was “to remove bad-faith actors and to advance public understanding of these critical topics.” Pickles defined state-backed information operations as “coordinated platform manipulation efforts that can be attributed with a high degree of confidence to state-affiliated actors.”
As the platform developed a more robust policy to deal with foreign influence operations and platform manipulation, it realized the Defense Department’s operations on their platform ran much deeper than the DOD’s initial whitelist requests.
According to emails reviewed by The Intercept, in May 2020, Lisa Roman, then Twitter’s VP of public policy, indicated that the platform had identified various accounts under DOD control that had not been disclosed to them in their initial request, and that the accounts “may violate our Rules.” In a separate email chain, Twitter head of trust and safety Yoel Roth stated that while the DOD expressed a desire to “work with us to remove the activity,” it was “refusing to discuss additional details or steps outside of a classified conversation.” A member of Twitter’s legal team indicated that the move to classify their activity on the platform was likely an effort to shield the extent of the operation from public scrutiny. Despite a massive suspension of identified clandestine accounts by Twitter and Facebook in 2020, the accounts in CENTCOM’s initial 2017 request remain whitelisted.
The Intercept was not able to obtain details or materials related to what took place in a subsequent meeting between representatives from the DOD and Twitter.
Fang indicated that while he was not granted unfettered access to internal files by Twitter, he was allowed over a period of three days to “make requests without restriction that were then fulfilled on my behalf by an attorney, meaning that the search results may not have been exhaustive.”