Empowering further research of potential information operations

Thursday, 31 January 2019

In October last year, we launched the first archive of all potential foreign information operations we have seen on Twitter. It is our fundamental belief that these accounts should be made public and searchable so members of the public, governments, and researchers can investigate, learn, and build media literacy capacities for the future.

So far, this archive of accounts and content has been accessed by thousands of researchers, governments, and people interested in learning more about foreign information operations. We have already seen valuable reporting and research produced on the basis of these datasets, including by major universities and think tanks in the United States and around the world.

Today, we are adding to this archive five new account sets we found based on continued contextual and semantic analysis from our investigations teams, one of the core components in our effort to protect the integrity of our service.

As noted last December, working with our industry peers we identified and suspended a very small number of accounts originating from Bangladesh for engaging in coordinated platform manipulation. The Tweets were entirely in Bengali and focused on regional political themes. All of these accounts and content are now part of the archive and can be investigated and reviewed by interested parties.

This post is unavailable
This post is unavailable.
This post is unavailable.

In August 2018, we were notified by an industry peer about possible malicious activity on their service about an attempted influence campaign we identified as potentially located within Iran. After receiving information from them, we began an investigation on our service to build our understanding of these networks. As soon as we discovered malicious activity, we immediately notified law enforcement and updated the public.


This post is unavailable
This post is unavailable.
This post is unavailable.

Since that time, through ongoing data science analysis, we have identified and suspended 2,617 additional malicious accounts that we believe may have origins in Iran. Twitter has been in close contact with our industry peers on this matter and shared detailed information with them about the malicious accounts. This multilateral process of information-sharing will continue, enabling us and our industry peers to work together to better understand and identify malicious activity.

In September 2018, Jack Dorsey testified on Capitol Hill on recent activities affiliated with the Russian Internet Research Agency (IRA), disclosing that Twitter had suspended a total of 3,841 accounts. Through ongoing analyses and investigations, we continue to build on our contextual understanding of these networks of accounts to improve our ability to find and suspend this activity as quickly as possible in the future. This is particularly vital as groups such as the IRA evolve their practices in response to industry-wide suspension efforts.

Our ongoing efforts have uncovered an additional 418 accounts. We cannot render definitive attribution to the IRA for these accounts, although most appear to originate in Russia, and much of the behavior mimics the activity of prior accounts tied to the IRA. We encourage researchers and other experts to study them and to contribute to general public awareness of these activities online.

Twitter has removed 764 accounts located in Venezuela. We are unable to definitively tie the accounts located in Venezuela to information operations of a foreign government against another country. However, these accounts are another example of a foreign campaign of spammy content focused on divisive political themes, and the behavior we uncovered is similar to that utilized by potential Russian IRA accounts. We are disclosing them out of an abundance of caution and welcome the feedback of researchers.

Additionally, we have removed 1,196 accounts located in Venezuela which appear to be engaged in a state-backed influence campaign targeting domestic audiences. We have shared information on these accounts with our industry peers, and continue to investigate malicious activity originating in Venezuela, both targeting audiences with in Venezuela and abroad.

What’s next?
As we have said, these issues are not new. Manipulation of information for national or geopolitical ends is part of human history and transcends ideological viewpoints. The medium of communication is what has changed. The behavior is against our values as a company. For our part, we are learning, evolving, and building a technological and personnel-driven approach to combating it. We hope that holistic, transparent disclosures such as this can help us all learn and build the necessary societal defenses and capacities to protect public conversation.


This post is unavailable
This post is unavailable.