Fighting disinfo with tech: A look at 3 start-ups

New technologies have made fake news and hostile disinformation more dangerous than ever: they spread like wildfire, supported by sophisticated networks of distribution like Facebook. Technology gave us this problem, could tech now be a key solution for the side of truth? We reached out to a couple of startups aiming to fight the fake news problem.

Factmata

Founded in 2017 in London, Factmata has already managed to attract more than $1 million of investments from figures such as Twitter co-founder Biz Stone or tech philanthropist Craig Newmark. 

Currently, Factmata develops three products, built around the need to automate the processing of online content. A scoring API takes text as input, and grades it on nine scales, including political bias, toxicity, hate speech and racism. This API uses natural language processing (NLP) in the background, and lets developers build Twitter bots to analyze news feeds or incorporate the technology in their apps – like news applications for example. The API comes in free and paid flavours, the latter raising the limitations present in the free edition.

The second product, “intelligence reports,” is aimed at analysing different claims and arguments around any topic. It goes through various related news articles, social media posts and other publications and then provides users with such insights as who are the main opinion-shapers or what is the general stance on the chosen topic. 

The last tool developed by Factmata is Moderation. This provides access to a constantly updated database of harmful websites that produce abusive or fake content. By relying on this dataset, advertisers can exclude these publishers from advertising networks and make sure their brand is not associated with fake news or disturbing content. The Factmata Moderation uses the same underlying technology as the previous two products to accurately classify media outlets as problematic and help advertisers avoid suspicious sites. The blacklist contains more than 10000 domains and is constantly updated.

Dhruv Ghulati, Factmata co-founder and CEO, tells TheFix that the company’s general goal is to create a score on any piece of content and tell how good or bad it is. “It’s a subjective thing so we will be trying to get it as objective as possible […] We want to change how the internet works,” he says.

Recently, Factmata got a significant investment from eyeo, the company which is famous for its Adblock Plus browser extension. As part of the deal, the Factmata team will lead the work on a newer browser extension called Trusted News, which was launched by eyeo last year. Trusted News warns internet users when they visit a problematic site, so they can take caution. Dhruv Ghulati believes it will help the startup attract new users. 

Even though  Factmata relies heavily on artificial intelligence (AI) in solving the problem of fake news, it doesn’t present itself as a tech company. The company claims to partner with more than two thousand experts, who help Factmata train its algorithm and provide more accurate results across all three product lines.

Nevertheless, Dhruv Ghulati believes that technology is increasingly important in the fight against disinformation. The technology’s growing role is informed by the necessity to scale the fact-checking process. “There is too much content […] We need to have an automated solution. Just like the mechanism to produce fake content is automated, the mechanisms to detect that content needs to be automated,” he says. 

Fakeskiller

A new promising Tallinn-based startup offers two main products — a browser plugin and a mobile app (the latter currently available only in Ukraine). The plugin analyses social media news feeds and flags known fake stories by matching them to the Fakeskiller database. The app was launched before the Ukrainian parliamentary elections, which were held in July. The idea behind the app is to give users the ability to check information about different political parties involved in the elections and show if they are spreading disinformation in its campaign. 

The Fakeskiller plugin connects the customers and the external experts working on the actual fact-checking. Currently, the plugin works only with social media but, according to the company’s CEO Dmytro Potekhin, they plan to start work with all webpages soon.

Startup accelerator Startup Wise Guys evaluated the company at 1 million euros. Potekhin says they are planning to find an additional investor in the nearest future. To make money, Fakeskiller plans to provide fact-checking services to companies suffering from reputation attacks. 

In an interview with TheFix, Dmytro Potekhin details his view of the place of technology in a big fight against disinformation. He believes that technology can be very helpful in making fact-checking faster and more accessible, especially the distribution of its results. But, in the end, it is people, not technology, who have the final impact. “I’m absolutely determined that the success in the fight against disinformation could be achieved only by social changes and the human factor,” he argues. Current technology is able to conduct a semantic analysis of text and detecting inappropriate vocabulary or hate speech, but it cannot really explain why the story is fake.

Logically

Logically is another London-based startup with the ambition to supply a tech-based solution against disinformation and improve civic discourse. Founded in 2017, the company has one main product, an app, which is currently available in the UK and India. Fact-checking is just one part of the product, Logically also wants their users to be able to “consume, discuss and verify”. 

That’s why the company also created a special discussion feature, which aims to be more than just a comments section below a news article. Its main idea is to give readers a “live-moderated environment” for discussing the news with friends or just random strangers. Users can verify different statements directly in the chat room and even kick out a user who violated the rules of by spreading fakes or showing abusive behaviour.

Logically presents itself to be a place for “intelligent” news consumption. The startup’s founder and CEO Lyric Jain tells TheFix they wanted to create the ”home for news in the digital world”. They want to create a place where users can consume news, pay attention to all sides of the story and have all the facts and perspectives in one place. 

Logically approaches fact-checking with an “on-demand” way. Users who consume news and have doubts about an article can file a claim, which Logically will then investigate. In terms of technology, Logically rellies both on AI and its human fact-checking team. Its AI technology uses a three-pronged approach — first, it analyses the metadata attached to the news item, such as who is the publisher and which site the content is coming from. Next, it examines the content itself. For that, Logically uses an NLP-based approach. And on the last phase, it looks at the article’s distribution pattern, mapping where it originates from and who amplifies it. After the claim has gone through this process, it goes to the human-fact checker team (around 25 people) which works on verification. On average, Logically receives around 500 claims per day.

Lyric says the app will be launched in the United States later this year, and it plans to expand further; particularly, to Eastern Europe. In terms of monetization, the app relies on the paid-subscription model. It has both free and premium versions available.

According to Lyric Jain, despite the increased importance of tech in fighting fake news, it’s not going to be the only efficient solution. “[The role of technology] is enabling fact-checking rather than being a single silver bullet” he argues. He refers to the concept of “extended intelligence” which they use in Logically. Instead of believing that AI can solve all the problems, they are looking at how we can use all the benefits of automation, as well as all the benefits of human intellect and creativity. 

“What can automation and tech do well? Scale, computation, looking at a lot of sources really quickly. What can a human do well? Humans can reason at a much more complicated level. Humans are much more flexible than any piece of tech.”

Lyric Jain, Logically CEO

Everybody seems to agree that the fake news problem lies in the massive scale of modern social networks. These networks are easy to game by less-than-honest players to amplify disinformation and fake news, spreading uncontrollably across billion-people nations like India and across the globe. Human-scale solutions simply can’t compete with the sheer size of these networks – automated solutions, complete with state-of-the-art AI in the form of NLP however can make a meaningful contribution in stopping fake news. Or at least we hope so.