This forensic thread details 🧵👇
- HOW it was done
- HOW it spread
- WHO runs them
1. HOW WAS THIS OPERATION CARRIED OUT?
Yesterday morning I came across these tweets, and was sent these WhatsApp messages from contacts in India.
What are they? They're links to a Google Drive, containing suggested tweets and graphics to use. Let's have a closer peek.
Here is the Google Drive with a Google Document containing text packets of suggested tweets, and a folder with images for the network to pump out through Twitter.
For a little context of this material, there are two main parties I refer to that are at heads with each other in the leadup to the #WestBengalElections2021. They are:
- The Bharatiya Janata Party (BJP), major Modi-led political party.
- The All India Trinamool Congress (TMC).
Here are the suggested tweets. They're in Bangla, English and Hindi. Many of them promote @AmitShah for his West Bengal tour as well as the BJP, some also discredit the opposing TMC party.
One text even claimed the opposition used a 'chant' to 'cheat' the people - Ironic.
It's not the first time we've seen this technique used for pro-BJP/Modi Govt content in India. Using a Google Doc appears to be a standard procedure to influence social media narratives.
Here's an example of one from last month also targeting Bengal 👇
This influence operation runs smoothly. Accounts take the text from the document, and tweet it out into their communities under a targeted hashtag.
Let's follow some of that text as it leaves the doc, and look at the accounts posting these packets.
The accounts posting the text are fairly obvious in their allegiances. For example, here's a selection of five accounts.
Three are BJP-affiliated groups. Two analysed using accountanalysis.app. show the past 1000 tweets of those accounts have majority pro-BJP/Modi content.
I also collected 5000+ accounts using the #AmitShahInBengal hashtag at the time the information operation was in effect.
There were a lot with twitter handles starting with the letters "bjp"
Here are also accounts in the #AmitShahInBengal network with "BJP" in their profile description.
3. DID THIS NETWORK HAVE ANY IMPACT?
Well first, let's take a look at the status of the #AmitShahInBengal tag.
In a 24 hour period it reached No.6 of most used tags in India.
That's effective. But that could also be due to normal user activity as well.
What about looking at the traction gained from the text taken from the Google Document that the influence operation was being ran from?
To show this, I've taken screenshots of Google Doc, and screenshots of the relevant copy&paste text used on Twitter.
We can see two things here.
1. There are verified accounts and accounts with good standing that have been part of this copy paste network.
2. There is an odd ratio of retweets to likes with some of these accounts getting 300+ retweets, yet less than 50 likes.
Some might ask why is this bad?
This was a trending topic in India yesterday. People were talking about it. However it's likely that many were unaware there was a coordinated attempt to push a specific agenda and discredit others in that dialogue.
That's inauthentic.
A goal of the campaign was to boost BJP profiles. So let's look at who is at the center of the network. We can visualise the entire network in @Gephi using data from Twitter's API to identify the accounts amplified.
Zooming in a little closer, we can see the main actors with their green, purple, pink and blue amplifier networks.
Those accounts in the center of the #AmitShahInBengal network are @AmitShah@BJP4India@BJP4Bengal and @KailashOnline.
By no means am I attributing the cause of the network to those accounts in the visualisation above. They are merely the central nodes in the network's activity.
4. WHO IS BEHIND NETWORKS LIKE THIS?
This is not the first pro-BJP operation to be waged on social media, but it does indicate common TTPs (Tactics, Techniques and Procedures) and there are digital crumbs left behind to identify those parties.
Let's look at some of the cases
CASE 1: In almost the exact same example of pro-BJP content in Bengal, again spread from a Google Doc, last month I identified a network targeting the #YuvaShaktiWithModi tag.
AUTHOR of the doc: a social media team leader for the Government 👇
CASE 2: Case from 2018 that saw the amplification of the tag #MamataKilledStudents. A screenshot of the Google Drive and Google Document, with suggested tweets (just like the current case) can be seen below.
AUTHOR of the Drive: seen in the first image as "BJP West Bengal"
CASE 3: Again very similar to all of these is a campaign to promote the tag #ModiWithFarmers.
AUTHOR: In the image below of that Google Drive, we can see the owner is labelled as ‘BJP IT & SM Office’ – this stands for BJP Information Technology and Social Media Office.
There are policies covering the issue of what I have documented on the repeated posting of identical tweets and using a hashtag to manipulate a conversation
Also an important point. It's not just Australia. The restriction is international, on Australian news publishers.
"Globally, posting and sharing news links from Australian publishers is also restricted".
That will be crippling.
Oh no! They got the Albany Advertiser Facebook page!
I did my cadetship here in 2009. A great community-based newspaper. It's just one of the many Facebook pages that have become a staple for media in Australia that are now cleared.
The copy and paste (copypasta) network is getting its text packets from this Google document. The tweets are in English, Hindi and Bangla.
Thank you to those who sent this through to me. It's an effective campaign to influence the topic of the coming elections.
The size of this sort of influence network? Well... I've just been collecting tweets under tag #BJPGorbeSonarBangla over the past 15 minutes. So far I've raked in 1789 tweets and growing 😅
Here's a visualisation of the main strand. The large nodes are: @AmitShah and @BJP4India
A copypasta campaign in India is targeting politically-sensitive tags #TMCHataoBanglaBachao & #KrishokSurokhaAbhijan on Twitter. I captured data over the past week for analysis.
It's clear there is coordinated activity here. Scores of pro-BJP/Pro-Modi accounts operate in a network to spam packets of text (copypasta method)
Example: the tag #TMCHataoBanglaBachao with the text “Bengal has been suffering from last 10 years. The TMC regime has built..."
And another related campaign, using the tag #KrishokSurokhaAbhijan with the text: "নতুন কৃষি আইনের দ্বারা কৃষিজ পণ্য বিক্রি করা হবে আরও সহজ। বিজেপি সরকার আনছে কৃষক সুরক্ষা অভিযান।"
Hi Twitter, having spent a lot of time on @googleearth, I've built up a collection art visible from space. I'd like to share some of my favs with you all in this thread 🛰️🎨
First up is the 'grandma' from Ella & Pitr in Pavilion 3, France, located here: goo.gl/maps/KGANS9Cge….
Exhibit 2 is also from Ella & Pitr, this time in Panaji, Goa, India. This one is of a swimmer, taking a rest on the roof of a carpark. It is located here: goo.gl/maps/Wpue1tDEN…. And you can find some of the other ones they did nearby here: xxlcollective.com/ella-pitr1
Exhibit 3 (one of my favs) is Art Attack from Kitt Bennett in Reservoir, Victoria, Australia (here: goo.gl/maps/nc1CbxY8E…). It is spread over three basketball courts. (original post here: instagram.com/p/Bc4AQUiBcmQ/)
A small network of fake accounts linked to an Iranian influence operation was posting about COVID & US Politics using accounts with bios & pics copied from UK media and activists & others disguised as 'fanatics'.
First, this investigation originated out of Twitter's data from a influence network takedown in October (screenshot is of data in that report): transparency.twitter.com/en/reports/inf…. The new accounts are the ones that survived.
Second, this network was small and had minimal impact.
I looked a little deeper into the accounts in the takedown report and started to find some strange things. Under many of the suspended accounts, comments had been made using the same text with the same video - so let's look at who's posting them.
Many of the accounts across Twitter, Facebook and Instagram used artificially-generated profile pictures (GAN) to disguise themselves. Here's an example from the accounts on Twitter - the detail is in the eyes 👀
Of course, if you're an influence researcher (or a news fanatic) this isn't the first time you would have seen fake accounts utilising GAN images. You might have seen examples from @elisethoma5 & @benimmo on them.
Here: