Connect with us

Support The NewsHawks

Obituary

Age of Disinformation: Building a next level bot to subvert Africa’s elections

Published

on

AMANDA SPERBER/JUSTIN ARENSTEIN

A six-month undercover probe using extensive analysis of digital evidence to investigate the secretive global disinformation-for-hire industry has unmasked a crusading investigative reporter, Anita Pettit, as a completely fictitious persona created by a shadowy Israeli private intelligence company, Percepto International, to smear opponents and subvert elections.

FOR crusading investigative reporter Anita Pettit, “the truth has no secrets”. A French-Ghanaian graduate of the Université Paris-Est Créteil, Pettit built a reputation for exposing injustice and what she calls “betrayal”.

Her hard-hitting opinionated reportage is published both on her own investigative website, Pour La Verité (“For The Truth”), and in African mass media, including La Revue de l’Afrique and Net Afrique and similar regional outlets, reaching a combined audience of over 10.2 million people, where she tried to shine a light on supposed crooked politicians and their links to terrorism and suspected corruption across eight African countries.

But the earnest pan-Africanist – and her portfolio of hundreds of exposés – is not what she seems. A six-month undercover probe by the #StoryKillers consortium of 30 newsrooms that used extensive analysis of digital evidence to investigate the secretive global disinformation-for-hire industry has unmasked Pettit as a completely fictitious persona, or “deep avatar”, created by a shadowy Israeli private intelligence company, Percepto International, to smear opponents and subvert elections.

Percepto later claimed to have used the Pettit sock-puppet and similar bots, as well as “deep platforms” such as fake media outlets or fake civil society groups, to secretly rig elections or smear major organisations such as the International Committee of the Red Cross (ICRC) in Burkina Faso, where Percepto framed it for supposedly working with jihadist terrorists (see story below).

“We’re currently operating dozens of assets online, unattributed assets in Africa, in Francophone Africa… our secret sauce is the use of avatars for intelligence gathering… [our] online assets are able to communicate, to engage, with targets,” Percepto’s co-founder and CEO Royi Burstien boasted during a series of secretly recorded meetings with the consortium’s undercover reporters.

“A deep avatar,” he explained, “is someone, an online entity, who looks real [but] that is really not true. And, if one of our deep avatars engages with you, you’ll be certain that [it’s] a real person. We have been operating this specific investigative journalist for the past, uh, from 2019. She is French. She lives in Paris, [but] she’s an avatar of course… 

“She’s already worked in three different countries, on three different projects. One project in France and two projects in two different countries in Africa. One Francophone, one Anglophone.”

Burstien claims Pettit was so effective she triggered retaliation.

“She published a piece attacking one of our clients’ opponents and we probably struck a nerve, because I think 24 hours after she published the piece, we had a cyber DDOS attack on [her] website. This was also picked up by a lot of cyber experts, and it gained huge traction on

Twitter in France and in specific countries. [O]ver the past four years, she engaged with a lot of journalists, including from France, including, by the way, a lot of French-speaking journalists in the diaspora outside of France.”

Percepto offered to make Pettit available as a tool for our undercover reporters, who he thought were representing a potential major new client in Africa that wanted to undermine a commercial competitor. 

The journalists were actually part of an initiative launched by the European non-profit Forbidden Stories, alongside major investigative media such as the GuardianLe MondeHaaretzDer Spiegel and 26 others, to expose the inner workings of the global world of disinformation mercenaries. Code for Africa (CfA) is the only African member of the consortium. 

Percepto offered to deploy the Pettit bot alongside other avatars that could help amplify whatever Pettit published.

Pettit’s digital backstory is elaborate, starting with heartfelt blogging about reconnecting with her African roots on her personal Ghanaland journal. She writes about busking in restaurants to save money to travel back to Ghana because, “I was born in France and I have a Ghanaian mother and a French father. Unfortunately, I did not really know my mom, who disappeared when I was very young. It’s hard to grow up without a mother by my side.”

Pettit writes that she’d been to Africa before, on a trip to visit her friend Sonia in Cameroon, but her “head and heart are with Ghana.” 

Travelling to Ghana, Pettit was a little nervous because English isn’t her first language, and she had some concerns about poverty and security, but after a long trip (she hates when the layover is longer than the flight) she arrived in Accra and was immediately struck by the hospitality. 

She appreciated that Sonia’s cousin, Souleman, welcomed her at the airport and helped her get to her hostel, the Sleepy Hippo, a well-reviewed, pet-friendly establishment with a pool on Duade Avenue in Kokomlemle, near the city centre. 

Pettit enjoyed her time in the country, visiting Kejetia market to buy “various handicrafts, souvenirs from all over the country, glass beads, all kinds of wooden sculptures, batik fabrics and many other wonderful things”, and travelling out of the capital to visit Kumasi and Cape Coast. 

Along the way, Pettit enjoyed a beer on the beach, until she spotted a boy who couldn’t be more than 12, drinking beer. Pettit was surprised and upset; shocked, even. 

“If I just tasted beer at this age, my father would have killed me,” she wrote. But, she came to understand, “More I look at them, more I understand that they have to live by themselves. No law and no limit for them. Just one goal: live as they can.”

But none of this is real: not her dreams and hopes, not the revelations she’s had along the way on her journey, and definitely not her supposed belief in the truth. 

The vast majority of the content on her investigative website is plagiarised from elsewhere, including leading French media such as Le Monde and Le Croix. The images on her personal blog from her travels are also edited versions of stock images from the Internet. Not even the profile photos on her social media on Facebook, Instagram and Twitter are real. 

Percepto constructed Pettit by stitching together stolen photos from at least three different people. A reverse image search of her Facebook profile picture shows that the picture had similarities with one posted on Instagram by a model based in Los Angeles, Sydney Graham. 

Pettit’s pose, clothing and the physical location are identical, but there is one big difference: the face. The facial features of the image have been changed and manipulated, presumably in an effort to evade detection or recognition by the original user.

On her Twitter feed, Pettit wishes herself happy birthday, using another “body-double” cloned from another model, this time a Paris-based model, Sharon Alexie. It was modified and posted as Pettit, using a different face, along with adjustments to make the body colour slightly darker in tone.

A reverse search of the face used on Anita’s profile shows that the image originated from a Jamaica-born engineer, Sherry-Ann Wellington, currently working in Russia. In fact, most of the facial images used by Pettit originate from Wellington’s profile.

Graham said she had never given Percepto or anyone else permission to use her image, and was not aware there was a doppelganger online. It was, she said, “weird and misleading”. Neither Alexie nor Wellington responded to multiple requests for comment.

Petitt’s profiles are being systematically removed by social media platforms after the #StoryKillers consortium alerted them to Percepto’s activities. 

The takedowns include a network of 23 other similar avatars, including another supposed investigative journalist, Chloé Boyer, whose profile claimed they were based in Paris. Boyer’s image appears to have been stolen from an account on the Russian VK social media platform, claiming to be a user called Svetlana Kerdich, that posts porn and humour memes. The Kerdich profile image was in turn lifted from another VK profile for a real-world user, Anna Vihareva, based in the Siberian city of Tomsk.

Percepto also tried to obscure the origin of Pettit’s supposed travel photos on her personal blog, by cropping/cutting the original photos they lifted from the Internet, and then inverting the image so that it would not appear as a duplicate if anyone tried to Google or otherwise look for the image. The original images are, however, easy to trace using specialised anti-fraud tools.

But, why go to all the effort to create such elaborate backstories?

Burstien told #StoryKillers that social media platforms like Facebook use algorithmic defences to find and delete avatars that aren’t convincing enough. Percepto, he insisted, is one of the few companies to have created unique techniques for evading detection. 

“We have avatar management systems and we have operational IP infrastructure [to] operate in Africa; in Cameroon. Senegal. Burkina Faso. After the Trump 2016 campaign, that’s when everybody started talking about unattributable avatars. That’s when it came to light. We’ve been doing this since 2013, 2014. And I think we’re one of the few companies that really has the longest track record in operating… for almost now over 10 years.” 

Burstien, a former Israeli intelligence operative, was previously CEO of the Psy-Group, which worked with the Israeli state as well as clients across the Gulf region until it was closed in the wake of Robert Mueller’s investigations into Russian interference in the 2016 American presidential election. 

And, because Percepto has been doing this for so long, some of the more complex avatars have built up years of “track record” across a wide variety of assignments, just like real humans. 

Burstien uses the metaphor of wine to explain, saying: “The older the avatar, the better the avatar.” This, he adds, is why Percepto’s networks remain operational, even while those operated by countries like Russia are regularly detected and destroyed.

Percepto’s deep avatars don’t just publish propaganda or smear opponents. Burstein claims that the more advanced avatars are also used to recruit and manage unsuspecting humans: real-world journalists or political activists.

The #StoryKillers consortium filed a set of detailed written questions with Burstien and Percepto, focusing on their claims during the four meetings with our reporters. Percepto responded by saying that it does not disclose confidential information, and would neither confirm nor deny the identity of its clients. 

It responded to a list of five questions about the avatars, including whether Percepto had created the Pettit bot or alternatively met/interacted with a real person by that name, with a single-word answer to each question: “No.” 

When questioned specifically about the ICRC disinformation campaign in Burkina Faso, Percepto said “the content of your question is untrue, but we do not comment on alleged activities performed by the company”.

The use of fake persona on social media is widespread in the disinformation industry. Another Israeli company unmasked by the #StoryKillers consortium, Team Jorge (TJ), appears to operate sophisticated digital tools that control an army of 30,000 avatars, all with fake profiles on Twitter, Facebook, Gmail, Instagram and LinkedIn. 

Some even have Amazon accounts with credit cards, Bitcoin wallets and Airbnb accounts. TJ is also led by former intelligence officers. 

#StoryKiller partner newsroom, the Guardian, explains that TJ’s toolkit, marketed as the Advanced Impact Media Solutions, or AIMS, controls fake social media profiles on an industrial scale to spread client’s propaganda at speed.

Unlike Percepto’s more targeted avatars, the TJ bots are used for coordinated comments or mass sharing of content. The campaigns driven by the bots appear to be mostly commercial disputes in about 20 countries, including the UK, US, Canada, Germany, Switzerland, Greece, Panama, Senegal, Mexico, Morocco, India, the United Arab Emirates, Zimbabwe, Belarus and Ecuador.

Twitter has suspended 1,775 accounts in the wake of #StoryKillers’ coverage, but has declined to comment on questions from the consortium. Meta, the owner of Facebook, also took down AIMS-linked bots on its platform after reporters shared a sample of the fake accounts with the company.

A Meta spokesperson connected the AIMS bots to others that were linked in 2019 to another, now defunct, Israeli firm which it banned from the platform.

“This latest activity is an attempt by some of the same individuals to come back and we removed them for violating our policies,” the spokesperson said.

“The group’s latest activity appears to have centred around running fake petitions on the Internet or seeding fabricated stories in mainstream media outlets.”–Daily Maverick.

Advertisement




Popular